- Find a configuration
- RTX 4090D · 48 GB
RTX 4090D · 48 GB
Task · models with qualified results on this GPU
Text generation · Qualified results
No qualified results yet · Text generation · RTX 4090D · 48 GB
445 original reports are kept as evidence but do not qualify for comparison; each report states why. This does not show that the model or GPU cannot run it.
Measured alternatives on other GPUs
Measured on different GPUs with the same or smaller listed capacity. These are references for choosing hardware, not performance or compatibility evidence for your selected GPU.
Qwen3.8-27B-NVFP4
RadixArk · 2 tested configurations · 1 GPU
- RTX 5090 · 32 GBNVFP4 · 8K → 2K tokens
- Output throughput · includes prefill · 1 request
- 70.8tok/s
- Observed GPU occupancy
- 27.3GiB
- First token
- 633 ms
Test details
Total output tokens divided by the full request window, including prefill and queueing.
Occupancy includes engine buffers and reserved KV cache. It is not the minimum VRAM required by the model.
Reports combined counts reports, not inference requests. Request successes and failures are listed in each report.
- Engine
- vllm
- Protocol
- wbench-fixed-http-v1
- Dataset
- synthetic-tokenizer-v1:repeats=2
- Workload
- 8192in-2048out
Concurrency Mean throughput (tok/s) P95 TTFT Observed occupancy (GiB) GPU power (W) Reports combined View report C1 70.82 29.01 s 633 ms 27.29 405.5 1 bad294c0 ↗ 1 other test configurations on this GPU
- NVFP4 · 32K → 2K tokens
- Output throughput · includes prefill · 1 request
- 62.5tok/s
- Observed GPU occupancy
- 27.3GiB
- First token
- 3.6 s
Test details
Total output tokens divided by the full request window, including prefill and queueing.
Occupancy includes engine buffers and reserved KV cache. It is not the minimum VRAM required by the model.
Reports combined counts reports, not inference requests. Request successes and failures are listed in each report.
- Engine
- vllm
- Protocol
- wbench-fixed-http-v1
- Dataset
- synthetic-tokenizer-v1:repeats=2
- Workload
- 32768in-2048out
Concurrency Mean throughput (tok/s) P95 TTFT Observed occupancy (GiB) GPU power (W) Reports combined View report C1 62.48 32.79 s 3.6 s 27.29 433.3 1 746bbc39 ↗
Qwen3-4B-AWQ
Qwen · 3 tested configurations · 2 GPUs
- L4 · 24 GBAWQ · 8K → 2K tokens · No cache reuse
- Output throughput · includes prefill · 1 request
- 55.3tok/s
- Best measured · 4 concurrent requests
- 109tok/s
- Observed GPU occupancy
- 19.5GiB
- First token
- 1.6 s
Use this recipe · 4 concurrent requests ↗Evidence report and run availability ↗Same test on other GPUsTest details · 2 concurrency levels
Total output tokens divided by the full request window, including prefill and queueing.
Occupancy includes engine buffers and reserved KV cache. It is not the minimum VRAM required by the model.
Reports combined counts reports, not inference requests. Request successes and failures are listed in each report.
- Engine
- vllm · vLLM 0.29.0
- Protocol
- wbench-fixed-http-v3
- Dataset
- migration-analysis-v1:sha256=68590c248e43f32c04dfd8062464fd1e4856917795fe307225e5c6112a54fe94:cache=isolated:repeats=3
- Workload
- 8192in-2048out
Concurrency Mean throughput (tok/s) P95 TTFT Observed occupancy (GiB) GPU power (W) Reports combined View report C1 55.3 37.17 s 1.6 s 19.54 71.9 1 9301a9ba ↗ C4 109.43 78.88 s 3.1 s 19.55 71.9 1 9301a9ba ↗ 1 other test configurations on this GPU
- AWQ · 8K → 2K tokens · Prefix cache reuse
- Output throughput · includes prefill · 1 request
- 57.4tok/s
- Best measured · 4 concurrent requests
- 175tok/s
- Observed GPU occupancy
- 19.5GiB
- First token
- 60 ms
Use this recipe · 4 concurrent requests ↗Evidence report and run availability ↗Same test on other GPUsTest details · 2 concurrency levels
Total output tokens divided by the full request window, including prefill and queueing.
Occupancy includes engine buffers and reserved KV cache. It is not the minimum VRAM required by the model.
Reports combined counts reports, not inference requests. Request successes and failures are listed in each report.
- Engine
- vllm · vLLM 0.29.0
- Protocol
- wbench-fixed-http-v3
- Dataset
- migration-analysis-v1:sha256=68590c248e43f32c04dfd8062464fd1e4856917795fe307225e5c6112a54fe94:cache=prefix-reuse:repeats=3
- Workload
- 8192in-2048out
Concurrency Mean throughput (tok/s) P95 TTFT Observed occupancy (GiB) GPU power (W) Reports combined View report C1 57.39 35.7 s 60 ms 19.55 71.8 1 7e2cbfda ↗ C4 175.39 46.75 s 114 ms 19.55 71.9 1 7e2cbfda ↗
Show 1 more GPUs
- T4 · 16 GBAWQ · 8K → 2K tokens · No cache reuse
- Output throughput · includes prefill · 1 request
- 15.4tok/s
- Best measured · 4 concurrent requests
- 20.3tok/s
- Observed GPU occupancy
- 12.7GiB
- First token
- 28.5 s
Use this recipe · 4 concurrent requests ↗Evidence report and run availability ↗Same test on other GPUsTest details · 2 concurrency levels
Total output tokens divided by the full request window, including prefill and queueing.
Occupancy includes engine buffers and reserved KV cache. It is not the minimum VRAM required by the model.
Reports combined counts reports, not inference requests. Request successes and failures are listed in each report.
- Engine
- vllm · vLLM 0.29.0
- Protocol
- wbench-fixed-http-v3
- Dataset
- migration-analysis-v1:sha256=68590c248e43f32c04dfd8062464fd1e4856917795fe307225e5c6112a54fe94:cache=isolated:repeats=3
- Workload
- 8192in-2048out
Concurrency Mean throughput (tok/s) P95 TTFT Observed occupancy (GiB) GPU power (W) Reports combined View report C1 15.39 133.36 s 28.5 s 12.74 67.9 1 fcfe2d64 ↗ C4 20.31 468.13 s 53.1 s 12.74 67.5 1 fcfe2d64 ↗
Occupancy includes engine buffers and reserved KV cache. It is not the minimum VRAM required by the model.
Quality by dataset
| Model | Dataset | Format | Engine | Average score | Sample size | Runs |
|---|---|---|---|---|---|---|
| RadixArk/Qwen3.8-27B-NVFP4 | hle-test-self-judge-22ed307-v1-n5-mt16384-t0 | NVFP4 | vllm | 20% | 5 / 2500 · Subset only | 1 runs20% · 7bbadab5 ↗ |
| RadixArk/Qwen3.8-27B-NVFP4 | simpleqa-verified-eval-self-judge-0dc97e0-v1-n1000-mt16384-t0 | NVFP4 | vllm | 6.4% | 1000 / 1000 · Full dataset | 1 runs6.4% · 404eb015 ↗ |
| RadixArk/Qwen3.8-27B-NVFP4 | simpleqa-verified-eval-self-judge-0dc97e0-v1-n5-mt16384-t0 | NVFP4 | vllm | 0% | 5 / 1000 · Subset only | 1 runs0% · 328fd742 ↗ |
| RadixArk/Qwen3.8-27B-NVFP4 | ifbench-test-official-loose-1c40f0c-v1-n300-mt16384-t0 | NVFP4 | vllm | 67.67% | 300 / 300 · Full dataset | 1 runs67.67% · 02a43d63 ↗ |
| RadixArk/Qwen3.8-27B-NVFP4 | ifbench-test-official-loose-1c40f0c-v1-n3-mt16384-t0 | NVFP4 | vllm | 66.67% | 3 / 300 · Subset only | 1 runs66.67% · 20ef4b9f ↗ |
| RadixArk/Qwen3.8-27B-NVFP4 | aime2026-test-zero-shot-strict-v4-n30-mt16384-t0 | NVFP4 | vllm | 63.33% | 30 / 30 · Full dataset | 1 runs63.33% · 87acbd23 ↗ |
| RadixArk/Qwen3.8-27B-NVFP4 | mmlu-pro-test-zero-shot-strict-v4-n3-mt16384-t0 | NVFP4 | vllm | 66.67% | 3 / 12032 · Subset only | 1 runs66.67% · 1d2bdd31 ↗ |
Measurements without a comparable protocol · 917 rows
These measured rates remain visible, but are excluded from comparison curves because the protocol is unknown.
| Model | Format | Engine | Workload | Concurrency | Throughput | GPU power (W) |
|---|---|---|---|---|---|---|
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.09 tok/s | 251 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.17 tok/s | 262.49 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.2 tok/s | 259.91 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.2 tok/s | 258.68 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.21 tok/s | 258.88 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.24 tok/s | 261.07 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 16.27 tok/s | 258.29 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 1 | 22.47 tok/s | 335.22 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 1 | 22.51 tok/s | 329.21 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 32.77 tok/s | 238.03 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 32.8 tok/s | 247.48 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 32.87 tok/s | 246.53 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 32.94 tok/s | 244.41 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 1 | 37.2 tok/s | 269.12 |
| Qwen3-32B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 38.27 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 44.09 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 44.09 tok/s | — |
| Qwen3-32B-AWQ ↗ | AWQ | vLLM | 2K | 1 | 46.34 tok/s | 323.63 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 51 tok/s | 227.48 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 51.27 tok/s | 225.16 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 51.42 tok/s | 229.09 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 51.51 tok/s | 219.25 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 51.58 tok/s | 228.74 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 51.65 tok/s | 226.69 |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 59.01 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 59.14 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 59.16 tok/s | — |
| Qwen3-32B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 68.31 tok/s | — |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 73.01 tok/s | 193.84 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 1 | 73.36 tok/s | 193.18 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 73.4 tok/s | 194.95 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 73.66 tok/s | 198.29 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 73.83 tok/s | 196.81 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.1 tok/s | 197.73 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.14 tok/s | 200.2 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.15 tok/s | 196.81 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.22 tok/s | 198.86 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.28 tok/s | 199.45 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.29 tok/s | 199.01 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 74.3 tok/s | 192.1 |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 77.87 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 77.92 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 83.57 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 83.65 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 83.71 tok/s | — |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 85.71 tok/s | 262.3 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 85.79 tok/s | 263.43 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 85.9 tok/s | 257.47 |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 1 | 91.53 tok/s | 305.5 |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 95.08 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 95.08 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 95.17 tok/s | — |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 1 | 99.14 tok/s | 302.52 |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 102.83 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 102.84 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 102.85 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 103.36 tok/s | 184.14 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 103.88 tok/s | 188.49 |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 105.18 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 105.42 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 105.45 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 105.51 tok/s | 189.72 |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 106.01 tok/s | — |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 106.03 tok/s | 185.4 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 106.2 tok/s | 192.05 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 106.2 tok/s | 191.4 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 1 | 106.34 tok/s | 191.64 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 1 | 106.38 tok/s | 193.41 |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 106.56 tok/s | — |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 106.67 tok/s | 191.68 |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 106.73 tok/s | — |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 110.42 tok/s | 283.05 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 113.89 tok/s | 284.17 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 116.06 tok/s | 286.72 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 121.1 tok/s | 280.53 |
| Qwen3-32B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 125.49 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 130.15 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 130.28 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 130.45 tok/s | — |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 133.11 tok/s | 279.95 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 134.43 tok/s | 176.42 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 135.49 tok/s | 180.04 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 135.58 tok/s | 179.28 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 136.31 tok/s | 178.28 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 136.82 tok/s | 173.65 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 136.83 tok/s | 283.86 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 136.85 tok/s | 177.93 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 136.95 tok/s | 280.71 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 136.96 tok/s | 177.68 |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 137.74 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 138.01 tok/s | — |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 138.43 tok/s | 184.6 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 138.45 tok/s | 182.55 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 1 | 139.77 tok/s | 282.76 |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 141.38 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 141.51 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 141.66 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 143.79 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 144.16 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 144.19 tok/s | — |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 8 | 156.01 tok/s | 329.45 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 158.85 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 159.16 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 159.55 tok/s | — |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 8 | 159.82 tok/s | 331.5 |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 161.17 tok/s | — |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 161.7 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 162 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 162.04 tok/s | — |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 162.73 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 164.97 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 165.13 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 165.14 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 170.23 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 170.27 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 170.39 tok/s | — |
| Qwen3-32B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 170.59 tok/s | — |
| Qwen3-32B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 174.51 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 175.17 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 175.36 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 175.67 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 175.89 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 176.07 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 179.71 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 179.94 tok/s | — |
| Qwen3-32B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 180.01 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 180.67 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 181.55 tok/s | 279.15 |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 181.77 tok/s | — |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 182.28 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 182.38 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 182.58 tok/s | 277.89 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 183.12 tok/s | 284.63 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 183.24 tok/s | — |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 186.16 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 186.28 tok/s | 281.54 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 186.29 tok/s | — |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 1 | 196.71 tok/s | 278.09 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 198.1 tok/s | 278.39 |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 200.4 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 202.4 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 202.73 tok/s | — |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 1 | 209.72 tok/s | 253.3 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 213.03 tok/s | 277.63 |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 213.5 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 216.5 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 219.72 tok/s | — |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 223.92 tok/s | 261.52 |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 224.67 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 224.88 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 1 | 225.21 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 227.39 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 228.11 tok/s | — |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 228.61 tok/s | 145.63 |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 228.82 tok/s | — |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 229.63 tok/s | 146.12 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 229.88 tok/s | 267.57 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 230.12 tok/s | 143.62 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 230.43 tok/s | 140.25 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 231.11 tok/s | 144.22 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 239.02 tok/s | — |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 239.71 tok/s | 265.89 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 240.92 tok/s | — |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 241.13 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 245.88 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 246.18 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 246.24 tok/s | — |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 248.37 tok/s | 264.54 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 249.24 tok/s | 211.56 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 249.26 tok/s | 209.3 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 1 | 249.58 tok/s | 208.11 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 250.76 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 252.85 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 252.98 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 259.11 tok/s | — |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 259.49 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 16K | 1 | 260.85 tok/s | 152.74 |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 261.63 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 261.78 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 262 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 262.31 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 262.95 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 265.72 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 1 | 267.49 tok/s | 146.51 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 32 | 267.6 tok/s | 381.6 |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 268.97 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 269.12 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 269.59 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 283.27 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 283.97 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 284.26 tok/s | — |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 16 | 285.03 tok/s | 353.6 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 16 | 290.43 tok/s | 356.86 |
| Qwen3-32B-AWQ ↗ | AWQ | vLLM | 2K | 8 | 292.18 tok/s | 315.33 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 297.97 tok/s | — |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 302.62 tok/s | — |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 303.71 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 313.55 tok/s | 243.16 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 317.91 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 318.22 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 319.52 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 319.79 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 320.03 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 2 | 320.07 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 326.17 tok/s | 240.98 |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 326.41 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 1 | 327.11 tok/s | 173.29 |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 328.42 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 328.96 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 329.71 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 8 | 331.25 tok/s | 228.1 |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 331.88 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 332.24 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 332.49 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 335.84 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 339.75 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 348.1 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 350.9 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 351.33 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 351.48 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 351.95 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 351.96 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 352.72 tok/s | — |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 354.67 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 355.07 tok/s | 245.27 |
| Qwen3-14B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 355.1 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 355.42 tok/s | 241.84 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 356.8 tok/s | 246.18 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 358.75 tok/s | 249.5 |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 2 | 358.96 tok/s | 184.1 |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 361.12 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 361.66 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 362.14 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 364.04 tok/s | 279.69 |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 365.02 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 365.53 tok/s | — |
| Qwen3-14B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 365.54 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 367.11 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 367.88 tok/s | — |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 1 | 368.05 tok/s | 220.2 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 1 | 368.57 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 369.93 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 370.11 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 370.35 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 370.9 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 371.07 tok/s | 284.36 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 372.39 tok/s | 280.54 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 373.01 tok/s | 262.2 |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 373.91 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 373.95 tok/s | — |
| Qwen3-14B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 374.13 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 378.01 tok/s | 284.42 |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 383.63 tok/s | — |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 387.07 tok/s | 248.16 |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 389.11 tok/s | — |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 393.18 tok/s | — |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 393.72 tok/s | 259.58 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 394.28 tok/s | 242.01 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 394.57 tok/s | 247.15 |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 396.76 tok/s | — |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 399.3 tok/s | 249.16 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 8 | 399.46 tok/s | 245.39 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 8 | 401.84 tok/s | 249.09 |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 402.53 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 406.49 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 406.64 tok/s | 282.73 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 406.92 tok/s | 245.31 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 407.24 tok/s | 263.18 |
| Qwen3-30B-A3B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 408.28 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 411.62 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 411.71 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 418.02 tok/s | — |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 418.77 tok/s | 260.19 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 420.66 tok/s | 246.98 |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 423.84 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 424.64 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 424.7 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 426.5 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 4 | 431.84 tok/s | 235.46 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 435.1 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 1 | 436.43 tok/s | 170.53 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 438.09 tok/s | — |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 438.17 tok/s | 278.61 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 438.26 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 439.37 tok/s | 206.75 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 443.99 tok/s | 286.02 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 445.88 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 447.2 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 1 | 447.36 tok/s | 174.14 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 447.85 tok/s | — |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 450.16 tok/s | 252.08 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 452.95 tok/s | — |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 453.56 tok/s | 251.87 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 454.5 tok/s | 249.55 |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 456.01 tok/s | — |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 456.96 tok/s | — |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 458.38 tok/s | 253.52 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 459.21 tok/s | 277.09 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 462.24 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 464.24 tok/s | 208.5 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 466.04 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 466.26 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 466.5 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 466.66 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 468.56 tok/s | 290.4 |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 469.7 tok/s | — |
| Qwen3-30B-A3B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 470.99 tok/s | — |
| Qwen3-32B-AWQ ↗ | AWQ | vLLM | 2K | 16 | 475.16 tok/s | 332.09 |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 486.39 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 487.05 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 487.93 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 488.85 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 489.7 tok/s | 229.33 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 32 | 492.31 tok/s | 391.28 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 494.33 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 494.42 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 2 | 495.99 tok/s | — |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 498.46 tok/s | 207.7 |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 499.3 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 510.31 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 519.12 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 526.72 tok/s | 211.09 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 8 | 531.12 tok/s | 212.55 |
| Gemma-3-270m ↗ | BF16 | vLLM | 32K | 1 | 532.47 tok/s | 189.12 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 532.7 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 533.33 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 534.4 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 536.37 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 32K | 1 | 536.84 tok/s | 190.17 |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 539.62 tok/s | — |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 541.59 tok/s | 210.75 |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 542.49 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 543.25 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 544.52 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 16K | 1 | 546.4 tok/s | 186.78 |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 549.67 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 550.14 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 550.2 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 550.33 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 550.36 tok/s | 259.99 |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 552.39 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 552.4 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 4 | 552.43 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 553.33 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 8K | 1 | 553.45 tok/s | 184.86 |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 553.91 tok/s | — |
| Qwen3-8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 554.33 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 555.43 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 555.94 tok/s | 263.88 |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 1 | 558.43 tok/s | 185.07 |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 560.16 tok/s | — |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 8 | 560.55 tok/s | 210.36 |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 560.68 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 1 | 562.52 tok/s | 185.58 |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 562.98 tok/s | — |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 564.72 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 565.03 tok/s | 186.89 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 565.23 tok/s | 190.9 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 1 | 565.84 tok/s | 184.72 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 1 | 566.15 tok/s | 185.97 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 1 | 566.41 tok/s | 184.31 |
| Qwen3-8B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 566.83 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 1 | 566.96 tok/s | 184.81 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 567.5 tok/s | 210.54 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 64 | 568.3 tok/s | 385.25 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 569.16 tok/s | — |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 8 | 569.24 tok/s | 214.92 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 569.79 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 569.89 tok/s | — |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 570.76 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 570.85 tok/s | — |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 571.37 tok/s | 291.69 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 1 | 571.89 tok/s | — |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 8 | 572.12 tok/s | 215.36 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 573.83 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 574.69 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 575.12 tok/s | — |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 8 | 576.77 tok/s | 296.35 |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 577.53 tok/s | — |
| Qwen3-8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 578.08 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 578.74 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 578.86 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 578.91 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 580.31 tok/s | — |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 580.67 tok/s | — |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 581.37 tok/s | — |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 582.69 tok/s | 254.22 |
| HY-MT1.5-7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 582.7 tok/s | — |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 64 | 583.13 tok/s | 382.87 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 583.48 tok/s | 257.81 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 256 | 591.39 tok/s | 386.76 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 591.65 tok/s | 257.54 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 591.72 tok/s | 258.46 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 256 | 594.19 tok/s | 392.55 |
| HY-MT1.5-7B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 595.63 tok/s | — |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 512 | 597.36 tok/s | 397.72 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 1 | 598.27 tok/s | 189.67 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 599.75 tok/s | 265.94 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 599.9 tok/s | 216.27 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 16 | 607.23 tok/s | 266.07 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 607.43 tok/s | 211.72 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 607.45 tok/s | 212.95 |
| HY-MT1.5-7B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 607.92 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 1 | 608.88 tok/s | 191.83 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 612.17 tok/s | 216.25 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 612.74 tok/s | 215.25 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 16 | 623.26 tok/s | 265.26 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 627.02 tok/s | 266.67 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 634.03 tok/s | 259.33 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 1 | 640.38 tok/s | 195.23 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 640.69 tok/s | 240.22 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 128 | 642.1 tok/s | 393.7 |
| Qwen2.5-72B-AWQ ↗ | AWQ | vLLM | 2K | 128 | 646.86 tok/s | 398.54 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 647.25 tok/s | 267.19 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 1 | 648.15 tok/s | 196.02 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 8 | 654.49 tok/s | 269.69 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 665.32 tok/s | 155.08 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 665.58 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 665.75 tok/s | 150.73 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 666.4 tok/s | 244.49 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 667.55 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 2 | 667.91 tok/s | — |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 676.99 tok/s | 287.63 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 1 | 681.68 tok/s | 200.79 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 1 | 682.05 tok/s | 200.71 |
| Qwen3-32B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 687.14 tok/s | 289.26 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 687.22 tok/s | 247.51 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 687.4 tok/s | 243.73 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 689.51 tok/s | 292.41 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 690.52 tok/s | 331.63 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 1 | 692.79 tok/s | 202.09 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 706.33 tok/s | 241.71 |
| Qwen3-32B-AWQ ↗ | AWQ | vLLM | 2K | 32 | 706.34 tok/s | 365.23 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 708.79 tok/s | 266.13 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 718 tok/s | 264.95 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 726.89 tok/s | 267.84 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 748.12 tok/s | 261.87 |
| Qwen2.5-0.5B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 1 | 749.41 tok/s | — |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 760.97 tok/s | 245.93 |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 762.92 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 763.08 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 764.16 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 764.89 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 765.63 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 766.16 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 767.09 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 8 | 767.56 tok/s | — |
| Qwen3-4B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 767.6 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 767.67 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 769.14 tok/s | — |
| Qwen3-4B-Q5_K_M ↗ | Q5_K_M | vLLM | 1K | 16 | 771.32 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 776.25 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 777.88 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 778.01 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 778.28 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 778.97 tok/s | — |
| Qwen3-4B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 780.32 tok/s | — |
| Qwen2.5-0.5B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 2 | 794.85 tok/s | — |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 801.87 tok/s | 346.85 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 803.48 tok/s | 218.36 |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 803.71 tok/s | 216.61 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 810.27 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 810.48 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 811.46 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 813.28 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 813.68 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 815.93 tok/s | — |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 16 | 825.69 tok/s | 212.27 |
| Gemma-3-270m ↗ | BF16 | vLLM | 32K | 2 | 827.48 tok/s | 173.12 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 838.91 tok/s | 261.45 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 847.58 tok/s | 331.13 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 847.59 tok/s | 210.24 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 855.11 tok/s | — |
| Gemma-3-270m ↗ | BF16 | vLLM | 16K | 2 | 856.9 tok/s | 169.84 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 860.65 tok/s | 261.52 |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 2 | 861.15 tok/s | 164.61 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 861.42 tok/s | — |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 865.73 tok/s | 209.24 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 866.94 tok/s | 252.35 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 871 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 4 | 872.37 tok/s | 243.84 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 874.73 tok/s | 201.65 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 882 tok/s | 262.59 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 8 | 883.62 tok/s | 273.55 |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 889.78 tok/s | 353.01 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 899.87 tok/s | — |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 901.53 tok/s | 206.79 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 901.77 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 902.61 tok/s | — |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 903.49 tok/s | 212.2 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 903.58 tok/s | 226.18 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 16 | 918.65 tok/s | 206.99 |
| Qwen2.5-0.5B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 4 | 922.16 tok/s | — |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 926.86 tok/s | 193.14 |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 16 | 931.96 tok/s | 301.54 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 16 | 933.97 tok/s | 208.24 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 16 | 935.67 tok/s | 212.32 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 942.71 tok/s | 262.69 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 954.81 tok/s | 193.76 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 967.34 tok/s | 196.55 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 968.24 tok/s | 252.7 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 968.6 tok/s | 265.2 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 968.67 tok/s | 205.21 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 972.07 tok/s | 247.73 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 975.24 tok/s | 268.89 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 975.63 tok/s | 194.74 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 32 | 980.52 tok/s | 268.33 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 994.12 tok/s | 268.01 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 32 | 996.01 tok/s | 267.35 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,006.86 tok/s | 207.2 |
| Qwen3-32B-AWQ ↗ | AWQ | vLLM | 2K | 64 | 1,028 tok/s | 369.03 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,050.25 tok/s | 196.18 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,051.26 tok/s | 196.45 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,051.67 tok/s | 199.69 |
| Qwen2.5-0.5B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 1,055 tok/s | — |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,058.96 tok/s | 201.86 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,060.97 tok/s | 201.32 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,063.11 tok/s | 212.6 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 1,099.69 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 1,100.18 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 1,103.61 tok/s | 160.07 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 1,104.22 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 4 | 1,105.47 tok/s | 157.62 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 16 | 1,107.11 tok/s | 263.53 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 1,138.09 tok/s | — |
| Qwen3-32B-AWQ ↗ | AWQ | vLLM | 2K | 128 | 1,160.53 tok/s | 369.83 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,162.31 tok/s | 248.22 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,179.37 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,180.43 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,183.28 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,185.95 tok/s | — |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,186.06 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,186.78 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,186.92 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,187.4 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,187.61 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 1,192.23 tok/s | — |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,193.6 tok/s | 233.56 |
| Qwen3-1.7B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,194.11 tok/s | 237.08 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 1,201.19 tok/s | — |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 8 | 1,202.43 tok/s | 250.26 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 1,205.36 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 1,205.87 tok/s | 249.31 |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 1,206.15 tok/s | 258.49 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,213.14 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 1,215.96 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 1,216.08 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 1,218.67 tok/s | — |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 8 | 1,219.05 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 8 | 1,219.77 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 1,220.29 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 1,220.63 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 1,221.21 tok/s | — |
| HY-MT1.5-1.8B-Q6_K ↗ | Q6_K | vLLM | 1K | 16 | 1,221.48 tok/s | 256.97 |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 1,221.5 tok/s | 249.53 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,221.79 tok/s | 275.3 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,229.45 tok/s | — |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,229.75 tok/s | 251.18 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,236.89 tok/s | 238.01 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,240.99 tok/s | 241.67 |
| HY-MT1.5-1.8B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,241.06 tok/s | — |
| Qwen3-VL-32B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 1,254.31 tok/s | 349.3 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,270.87 tok/s | 275.84 |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 8 | 1,272.44 tok/s | 253.07 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 16 | 1,273.62 tok/s | 269.7 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,294.65 tok/s | 274.85 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,301.63 tok/s | 252.53 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,302.67 tok/s | 272.62 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 8 | 1,339.85 tok/s | 185.83 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,342.16 tok/s | 275.16 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,352.57 tok/s | 277.38 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,355.61 tok/s | 274.82 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,361.86 tok/s | 251.1 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,362.38 tok/s | 273.38 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,372.82 tok/s | 276.92 |
| Qwen2.5-0.5B-Q4_K_M ↗ | Q4_K_M | vLLM | 1K | 16 | 1,378.74 tok/s | — |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,389.92 tok/s | 273.04 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,403.32 tok/s | 275.51 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,426.13 tok/s | 250.87 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,436.15 tok/s | 251.81 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 8 | 1,445.3 tok/s | 164.85 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,456.73 tok/s | 172.69 |
| Gemma-3-270m ↗ | BF16 | vLLM | 32K | 8 | 1,458.27 tok/s | 157.46 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,470.9 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,471.65 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,477.03 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,478.49 tok/s | 170.53 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,478.81 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,481.7 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,483.75 tok/s | — |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 16 | 1,490.8 tok/s | 170.77 |
| Qwen3-0.6B-Q8_0 ↗ | Q8_0 | vLLM | 1K | 8 | 1,490.92 tok/s | 169.76 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,516.6 tok/s | 275.01 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,519.97 tok/s | 214.53 |
| Gemma-3-270m ↗ | BF16 | vLLM | 32K | 4 | 1,522.33 tok/s | 192.63 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,527.27 tok/s | 211.02 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,534.73 tok/s | 214.07 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,535.83 tok/s | 223.71 |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 32 | 1,539.11 tok/s | 339.88 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,547.49 tok/s | 212.55 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 8 | 1,552.03 tok/s | 149.96 |
| Gemma-3-270m ↗ | BF16 | vLLM | 16K | 8 | 1,552.35 tok/s | 149.86 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,556.33 tok/s | 223.4 |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 4 | 1,559.22 tok/s | 178.04 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,562.54 tok/s | 189.88 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,574.68 tok/s | 221.34 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,575.67 tok/s | 186.03 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,580.25 tok/s | 171.1 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,590.69 tok/s | 221.52 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,590.87 tok/s | 170.77 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,591.62 tok/s | 171.52 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,592.69 tok/s | 171.19 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,596.36 tok/s | 273.4 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 8 | 1,598.01 tok/s | 174.09 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 8 | 1,599.27 tok/s | 149.04 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 8 | 1,602.47 tok/s | 142.08 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,607.47 tok/s | 218.44 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,611.31 tok/s | 188.05 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,612.34 tok/s | 190.12 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 32 | 1,618.13 tok/s | 281.54 |
| Gemma-3-270m ↗ | BF16 | vLLM | 16K | 4 | 1,630.3 tok/s | 187.19 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 8 | 1,632.06 tok/s | 141.7 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,638.15 tok/s | 212.69 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,645.04 tok/s | 218.77 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 8 | 1,649.66 tok/s | 141.18 |
| Gemma-3-270m ↗ | BF16 | vLLM | 8K | 8 | 1,651.81 tok/s | 146.62 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,665.06 tok/s | 223.52 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 8 | 1,665.74 tok/s | 141.35 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,678.88 tok/s | 216.37 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,682.08 tok/s | 266.2 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 8 | 1,687.01 tok/s | 140.83 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,687.45 tok/s | 264.22 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 8 | 1,695.08 tok/s | 138.95 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 8 | 1,699.55 tok/s | 158.78 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,709.98 tok/s | 266.64 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 8 | 1,713.17 tok/s | 140.12 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,738.28 tok/s | 191.06 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,738.4 tok/s | 189.7 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,739.89 tok/s | 190.26 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 8 | 1,747.17 tok/s | 143.73 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,750.12 tok/s | 196.14 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 1,752.59 tok/s | 192.7 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,828.23 tok/s | 219.23 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 8 | 1,846.76 tok/s | 138.58 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 1,847.98 tok/s | 224.12 |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 16 | 1,903.17 tok/s | 244.57 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 8 | 1,917.24 tok/s | 142.39 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 1,925.13 tok/s | 325.6 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 8 | 1,925.32 tok/s | 222.84 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 1,927.99 tok/s | 323.35 |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 16 | 1,931.19 tok/s | 236.68 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 1,937.23 tok/s | 325.71 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,963.24 tok/s | 262.14 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 1,975.17 tok/s | 262.92 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 1,990.19 tok/s | 325.32 |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 64 | 2,074.24 tok/s | 347.03 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,107.95 tok/s | 290.21 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,108.81 tok/s | 289.3 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,129.26 tok/s | 286.14 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,156.26 tok/s | 288.11 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,166.93 tok/s | 285.3 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,198.82 tok/s | 284.83 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 8 | 2,225.08 tok/s | 146.97 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,268.81 tok/s | 267.2 |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 128 | 2,286.57 tok/s | 344.42 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 32 | 2,299.42 tok/s | 303.56 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 8 | 2,330.32 tok/s | 153.08 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,330.53 tok/s | 265.76 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,334.15 tok/s | 296.53 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,351.02 tok/s | 269.89 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,369.35 tok/s | 267.96 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,461.2 tok/s | 287.36 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,472.09 tok/s | 285.22 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,477.56 tok/s | 233.58 |
| Gemma-3-270m ↗ | BF16 | vLLM | Mixed | 8 | 2,479.12 tok/s | 161.51 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,484.72 tok/s | 343.34 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,485.11 tok/s | 162.18 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,485.19 tok/s | 163.23 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,490.38 tok/s | 161.21 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,491.82 tok/s | 161.07 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,500.48 tok/s | 163.99 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,501.77 tok/s | 337.38 |
| Qwen3-14B-AWQ ↗ | AWQ | vLLM | 2K | 256 | 2,503.73 tok/s | 353.52 |
| Qwen3-14B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,516.02 tok/s | 336.15 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,534.64 tok/s | 232.52 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,554.3 tok/s | 233.73 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,554.99 tok/s | 305.69 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,562.45 tok/s | 240.53 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,566.37 tok/s | 238.09 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,590.63 tok/s | 204.2 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,596.37 tok/s | 302.9 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,599.07 tok/s | 305.71 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,611.51 tok/s | 304.15 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,623.08 tok/s | 207.92 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,628.95 tok/s | 206.37 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,636.18 tok/s | 205.06 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,636.56 tok/s | 242.16 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,653.11 tok/s | 239.51 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,654.38 tok/s | 293.61 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,665.72 tok/s | 241.49 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 8 | 2,667.02 tok/s | 161.76 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,668.28 tok/s | 295.58 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,669.39 tok/s | 242.11 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,671.61 tok/s | 226.59 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 16 | 2,677.74 tok/s | 201.02 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,679.18 tok/s | 209.15 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,690.33 tok/s | 238.98 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,690.73 tok/s | 292.87 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,695.85 tok/s | 231.18 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 16 | 2,706.58 tok/s | 228.46 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,768.2 tok/s | 302.62 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 2,800.56 tok/s | 301.17 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 32 | 2,851.12 tok/s | 205.79 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,851.94 tok/s | 285.61 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,879.43 tok/s | 288.82 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,885.55 tok/s | 240.16 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 2,888.89 tok/s | 239.39 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,896.36 tok/s | 214.16 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,899.87 tok/s | 209.93 |
| Qwen3-VL-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 2,901.88 tok/s | 286.26 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 256 | 2,903.16 tok/s | 294.06 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,903.83 tok/s | 212.69 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 2,903.95 tok/s | 213.06 |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 32 | 2,922.56 tok/s | 272.3 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 64 | 2,959.2 tok/s | 298.78 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 64 | 2,999.88 tok/s | 316.37 |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,003.25 tok/s | 292.33 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 128 | 3,016.02 tok/s | 316.05 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,049.61 tok/s | 277.57 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,075.68 tok/s | 280.2 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,095.75 tok/s | 277.3 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,096.26 tok/s | 312.51 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 128 | 3,097.45 tok/s | 343.47 |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 32 | 3,118 tok/s | 263.79 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,132.23 tok/s | 309.61 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,150.16 tok/s | 277.72 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 3,179.04 tok/s | 292.35 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,194.65 tok/s | 279.21 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 3,197.39 tok/s | 289.54 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,210.93 tok/s | 276.06 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,246.14 tok/s | 280.48 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,249.3 tok/s | 278.92 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,274.96 tok/s | 309.95 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,292.09 tok/s | 280.22 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,357.9 tok/s | 290.56 |
| Qwen3-8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,371.27 tok/s | 311.44 |
| Qwen3-30B-A3B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,384.65 tok/s | 290.25 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,404.39 tok/s | 273.65 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,406.27 tok/s | 311 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,411.53 tok/s | 307.65 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,419.49 tok/s | 270.99 |
| Qwen3-8B-AWQ ↗ | AWQ | vLLM | 2K | 256 | 3,432.32 tok/s | 314.78 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,436.23 tok/s | 272.94 |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 64 | 3,470.2 tok/s | 291.51 |
| Qwen3-VL-8B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 3,479.51 tok/s | 309.82 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 3,515.54 tok/s | 312.46 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 3,648.6 tok/s | 214.73 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 3,680.94 tok/s | 216.53 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 3,687.18 tok/s | 217.51 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 3,715.75 tok/s | 231.34 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 3,720.63 tok/s | 230.27 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 3,769.38 tok/s | 231.58 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 3,789.86 tok/s | 233.5 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,833.43 tok/s | 306.58 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 3,957.84 tok/s | 293.33 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 256 | 3,982.47 tok/s | 347.41 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,010.29 tok/s | 308.72 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,015.68 tok/s | 298.33 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,022.43 tok/s | 297.43 |
| Hunyuan-7B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,024.58 tok/s | 309.83 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,039.04 tok/s | 294.68 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,073.56 tok/s | 297.38 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,074.31 tok/s | 293.27 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,086.05 tok/s | 294.86 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 4,088.45 tok/s | 185.66 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 4,090.68 tok/s | 183.52 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 4,092.97 tok/s | 186.86 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 4,094.54 tok/s | 182.77 |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 128 | 4,099.79 tok/s | 304.82 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 32 | 4,118.01 tok/s | 187.59 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,126.67 tok/s | 292.11 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,139.24 tok/s | 292.68 |
| Qwen3-4B ↗ | BF16 | vLLM | 2K | 512 | 4,144.53 tok/s | 348.51 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 32 | 4,179.65 tok/s | 234.44 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,185.57 tok/s | 293.15 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,199.45 tok/s | 291.88 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,200.93 tok/s | 298.09 |
| Qwen3-4B-Thinking-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,205.28 tok/s | 297.69 |
| Qwen3-4B-AWQ ↗ | AWQ | vLLM | 2K | 256 | 4,223.48 tok/s | 302.05 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,223.99 tok/s | 298.36 |
| Qwen3-VL-4B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,231.81 tok/s | 292.9 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,248.29 tok/s | 297.86 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,266.23 tok/s | 292.16 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,279.09 tok/s | 293.24 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 4,280.04 tok/s | 242.05 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 4,280.71 tok/s | 239.22 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 4,281.05 tok/s | 236.55 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 4,281.55 tok/s | 238.65 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 4,282.26 tok/s | 236.48 |
| Qwen3-4B-Instruct-FP8 ↗ | FP8 | vLLM | 2K | 512 | 4,301.45 tok/s | 296.96 |
| Qwen3-4B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 4,510.88 tok/s | 292.8 |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 64 | 4,571.17 tok/s | 276.8 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 32 | 4,736.15 tok/s | 188.5 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,022.91 tok/s | 261.06 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,031.73 tok/s | 263.84 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 5,035.17 tok/s | 235.97 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,044.73 tok/s | 264.21 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 5,049.38 tok/s | 234.03 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,063.88 tok/s | 263.3 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 32 | 5,070.66 tok/s | 185.03 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 5,112.76 tok/s | 231.93 |
| Gemma-3-270m ↗ | BF16 | vLLM | 8K | 32 | 5,116.35 tok/s | 192.84 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 512 | 5,229.81 tok/s | 252.58 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 32 | 5,235.34 tok/s | 181.53 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,315.85 tok/s | 268.63 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,317.53 tok/s | 264.7 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,318.32 tok/s | 265.05 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,319.29 tok/s | 264.76 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 5,320.59 tok/s | 265.27 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 32 | 5,431.31 tok/s | 184.86 |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 128 | 5,461.19 tok/s | 303.63 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,471.64 tok/s | 277.59 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 32 | 5,480.34 tok/s | 181.78 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 32 | 5,509.22 tok/s | 179.5 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,519.06 tok/s | 279.59 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 32 | 5,519.73 tok/s | 177.48 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,523.21 tok/s | 277.02 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,534.9 tok/s | 276.54 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 32 | 5,586.19 tok/s | 178.08 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,618.03 tok/s | 278.51 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 32 | 5,676.02 tok/s | 176.12 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 64 | 5,676.24 tok/s | 261.53 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 5,709.94 tok/s | 277.06 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 5,738.93 tok/s | 278.59 |
| Qwen3-VL-2B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 5,754.84 tok/s | 276.92 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 5,783.67 tok/s | 280.65 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 32 | 5,806.54 tok/s | 172.88 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 32 | 5,829.92 tok/s | 171.14 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 5,969.07 tok/s | 285.44 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 5,973.7 tok/s | 280.9 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,987.98 tok/s | 282.56 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,992.59 tok/s | 279.28 |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 512 | 5,993.48 tok/s | 319.17 |
| Qwen3-1.7B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 5,995.28 tok/s | 282.88 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 6,017.48 tok/s | 250.6 |
| Qwen3-1.7B ↗ | BF16 | vLLM | 2K | 256 | 6,019.91 tok/s | 319.95 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 6,043.67 tok/s | 254.45 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 32 | 6,067.09 tok/s | 161.77 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 6,086.99 tok/s | 252.24 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 128 | 6,102.07 tok/s | 253.16 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 32 | 6,119.2 tok/s | 169.44 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 6,224.49 tok/s | 214.96 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 32 | 6,225.67 tok/s | 157.92 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 6,241.98 tok/s | 219.05 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 6,245.89 tok/s | 216.88 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 6,246.15 tok/s | 218.29 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 64 | 6,255.71 tok/s | 218.61 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 128 | 6,268.94 tok/s | 279.97 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 32 | 6,287.08 tok/s | 156.81 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 32 | 6,305 tok/s | 155.19 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 32 | 6,494.38 tok/s | 156.56 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 6,523.32 tok/s | 270.23 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 6,554.34 tok/s | 271.02 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 32 | 6,590.82 tok/s | 162.13 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 6,593.06 tok/s | 270.02 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 128 | 6,609.37 tok/s | 263.43 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 512 | 6,633 tok/s | 289.9 |
| Qwen3-0.6B ↗ | BF16 | vLLM | 2K | 256 | 6,682.18 tok/s | 290.79 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 6,772.9 tok/s | 269.82 |
| Hunyuan-1.8B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 6,777 tok/s | 271.56 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 128 | 7,607.28 tok/s | 243.69 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 512 | 7,885.21 tok/s | 266.09 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 8,087.53 tok/s | 256.21 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 8,088.78 tok/s | 262.62 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 8,090.46 tok/s | 255.69 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 8,109.16 tok/s | 260.79 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 128 | 8,111.09 tok/s | 260.13 |
| Gemma-3-270m ↗ | BF16 | vLLM | 8K | 512 | 8,452.13 tok/s | 267.55 |
| Gemma-3-270m ↗ | BF16 | vLLM | 8K | 128 | 8,584.2 tok/s | 252.28 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 128 | 8,913.21 tok/s | 243.45 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 512 | 8,939.01 tok/s | 261.82 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 128 | 8,967.4 tok/s | 240.86 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 128 | 9,108.9 tok/s | 248.31 |
| Gemma-3-270m ↗ | BF16 | vLLM | 4K | 512 | 9,211.88 tok/s | 261.53 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1024 | 9,270.05 tok/s | 290.67 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 1024 | 9,293.75 tok/s | 288.38 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 9,314.63 tok/s | 288.79 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 9,322.34 tok/s | 289.15 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 512 | 9,328.49 tok/s | 285.28 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 9,373.14 tok/s | 281.63 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 9,378.78 tok/s | 286.19 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 9,380.9 tok/s | 283.27 |
| Qwen3-0.6B-FP8 ↗ | FP8 | vLLM | 2K | 256 | 9,389.58 tok/s | 287.53 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 128 | 9,516.01 tok/s | 234.75 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 512 | 9,695.96 tok/s | 263.99 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 128 | 9,781.99 tok/s | 237.32 |
| Gemma-3-270m ↗ | BF16 | vLLM | 2K | 512 | 9,785.93 tok/s | 260.17 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 128 | 9,836.61 tok/s | 232.35 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 512 | 10,115.03 tok/s | 263.25 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 128 | 10,299.83 tok/s | 228.27 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 128 | 10,407.53 tok/s | 218.17 |
| Gemma-3-270m ↗ | BF16 | vLLM | 1K | 512 | 10,434.61 tok/s | 262.5 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 128 | 10,816.18 tok/s | 219.66 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 512 | 11,978.52 tok/s | 250.64 |
| Gemma-3-270m ↗ | BF16 | vLLM | 512 | 512 | 12,144.95 tok/s | 252.42 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 128 | 12,219.3 tok/s | 205.72 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 128 | 12,576.67 tok/s | 211.73 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 128 | 12,831.27 tok/s | 199.11 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 512 | 14,312.31 tok/s | 236.52 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 512 | 14,354.97 tok/s | 230.05 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 512 | 14,492.99 tok/s | 231.59 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 128 | 14,741.29 tok/s | 209.43 |
| Gemma-3-270m ↗ | BF16 | vLLM | 256 | 512 | 14,801.51 tok/s | 230.89 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 128 | 14,844.36 tok/s | 218.56 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 128 | 15,588.69 tok/s | 212.36 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 512 | 18,723.26 tok/s | 227.31 |
| Gemma-3-270m ↗ | BF16 | vLLM | 128 | 512 | 18,775.66 tok/s | 242.81 |
Original reports
| Model | Test task | Format | Engine | Workload | Protocol | Quality score | Test date |
|---|---|---|---|---|---|---|---|
| AutoGLM-Phone-9B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 11:34:43 AM |
| DASD-4B-Thinking ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 5:25:35 PM |
| DeepSeek-OCR-2 ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 7:31:44 AM |
| DeepSeek-R1-Distill-Llama-8B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 6:28:03 PM |
| DeepSeek-R1-Distill-Qwen-1.5B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 4:37:38 PM |
| DeepSeek-R1-Distill-Qwen-1.5B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 5:42:06 AM |
| DeepSeek-R1-Distill-Qwen-7B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 5:56:49 PM |
| dots.ocr ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 7:54:51 AM |
| gemma-3-270m ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 1/25/2026, 4:47:46 AM |
| gemma-3-270m ↗ | Text generation | Native | vLLM | 1K | Method unknown | — | 1/25/2026, 4:33:01 AM |
| gemma-3-270m ↗ | Text generation | Native | vLLM | 512 | Method unknown | — | 1/25/2026, 4:23:12 AM |
| gemma-3-270m ↗ | Text generation | Native | vLLM | 256 | Method unknown | — | 1/25/2026, 4:14:45 AM |
| gemma-3-270m ↗ | Text generation | Native | vLLM | 128 | Method unknown | — | 1/25/2026, 4:07:02 AM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | Mixed | Method unknown | — | 3/6/2026, 10:54:28 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128K | Method unknown | — | 3/6/2026, 10:51:17 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 64K | Method unknown | — | 3/6/2026, 10:47:54 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 32K | Method unknown | — | 3/6/2026, 10:43:08 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 16K | Method unknown | — | 3/6/2026, 10:39:20 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 8K | Method unknown | — | 3/6/2026, 10:29:22 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 4K | Method unknown | — | 3/6/2026, 10:21:36 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 2K | Method unknown | — | 3/6/2026, 10:15:08 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 1K | Method unknown | — | 3/6/2026, 10:08:36 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 512 | Method unknown | — | 3/6/2026, 10:01:36 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 256 | Method unknown | — | 3/6/2026, 9:50:42 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 3/6/2026, 9:25:21 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 3/6/2026, 9:06:52 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 8K | Method unknown | — | 3/6/2026, 9:01:54 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | Mixed | Method unknown | — | 3/6/2026, 8:55:56 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 4K | Method unknown | — | 3/6/2026, 8:55:28 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128K | Method unknown | — | 3/6/2026, 8:51:38 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 64K | Method unknown | — | 3/6/2026, 8:48:24 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 2K | Method unknown | — | 3/6/2026, 8:46:56 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 32K | Method unknown | — | 3/6/2026, 8:45:10 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 16K | Method unknown | — | 3/6/2026, 8:40:32 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 1K | Method unknown | — | 3/6/2026, 8:39:50 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 8K | Method unknown | — | 3/6/2026, 8:34:40 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 512 | Method unknown | — | 3/6/2026, 8:31:13 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 4K | Method unknown | — | 3/6/2026, 8:26:23 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 2K | Method unknown | — | 3/6/2026, 8:17:37 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 256 | Method unknown | — | 3/6/2026, 8:16:37 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 1K | Method unknown | — | 3/6/2026, 8:09:47 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 512 | Method unknown | — | 3/6/2026, 8:02:55 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 3/6/2026, 7:53:02 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 256 | Method unknown | — | 3/6/2026, 7:52:24 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 3/6/2026, 7:43:39 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 3/6/2026, 4:29:01 PM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 1K | Method unknown | — | 1/25/2026, 6:03:52 AM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 512 | Method unknown | — | 1/25/2026, 5:25:46 AM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 256 | Method unknown | — | 1/25/2026, 5:16:21 AM |
| Gemma-3-270m ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 1/25/2026, 5:07:39 AM |
| Hunyuan-1.8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 10:28:46 AM |
| Hunyuan-1.8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 1:05:02 AM |
| Hunyuan-1.8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 3:53:51 PM |
| Hunyuan-7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 11:14:03 AM |
| Hunyuan-7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 1:34:54 AM |
| Hunyuan-7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 6:58:29 PM |
| HunyuanOCR ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 6:57:44 AM |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/22/2026, 4:37:21 AM |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:33:27 PM |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:27:24 PM |
| HY-MT1.5-1.8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 3:32:41 PM |
| HY-MT1.5-1.8B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/22/2026, 4:39:13 AM |
| HY-MT1.5-1.8B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/20/2026, 11:35:20 PM |
| HY-MT1.5-1.8B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/20/2026, 7:29:13 PM |
| HY-MT1.5-1.8B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/20/2026, 3:34:32 PM |
| HY-MT1.5-1.8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/22/2026, 4:41:06 AM |
| HY-MT1.5-1.8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 11:37:10 PM |
| HY-MT1.5-1.8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:31:06 PM |
| HY-MT1.5-1.8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 3:36:24 PM |
| HY-MT1.5-7B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:44:39 PM |
| HY-MT1.5-7B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:38:44 PM |
| HY-MT1.5-7B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 3:44:03 PM |
| HY-MT1.5-7B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/20/2026, 11:46:34 PM |
| HY-MT1.5-7B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/20/2026, 7:40:39 PM |
| HY-MT1.5-7B-Q6_K ↗ | Text generation | Q6_K | vLLM | 1K | Method unknown | — | 1/20/2026, 3:46:40 PM |
| HY-MT1.5-7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 11:48:28 PM |
| HY-MT1.5-7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:42:35 PM |
| HY-MT1.5-7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 3:50:25 PM |
| LightOnOCR-2-1B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 6:35:52 AM |
| LightOnOCR-2-1B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 5:30:22 AM |
| MAI-UI-8B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 9:06:04 AM |
| MiniCPM-o-4_5 ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 9:54:41 AM |
| Nanbeige4-3B-Thinking-2511 ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 4:56:30 PM |
| PaddleOCR-VL-1.5 ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 6:12:32 AM |
| PaddleOCR-VL-1.5 ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 6:06:59 AM |
| PaddleOCR-VL-1.5 ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 4:34:39 AM |
| Qwen2.5-0.5B-Instruct-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 2/7/2026, 4:24:16 AM |
| Qwen2.5-0.5B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:24:33 AM |
| Qwen2.5-1.5B-Instruct-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 2/7/2026, 6:52:03 AM |
| Qwen2.5-1.5B-Instruct-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 2/7/2026, 6:46:46 AM |
| Qwen2.5-1.5B-Instruct-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 2/7/2026, 6:36:02 AM |
| Qwen2.5-1.5B-Instruct-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 2/7/2026, 6:24:22 AM |
| Qwen2.5-7B-Instruct-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 2/6/2026, 5:25:17 AM |
| Qwen2.5-72B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/14/2026, 8:15:40 AM |
| Qwen2.5-72B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/12/2026, 8:12:35 PM |
| Qwen3-0.6B ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 1/23/2026, 3:39:14 AM |
| Qwen3-0.6B ↗ | Text generation | BF16 | vLLM | 256 | Method unknown | — | 1/22/2026, 10:04:49 PM |
| Qwen3-0.6B ↗ | Text generation | BF16 | vLLM | 2K | Method unknown | — | 1/15/2026, 3:37:15 PM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/11/2026, 2:14:25 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 128 | Method unknown | — | 2/10/2026, 10:28:01 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 9:45:17 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/6/2026, 6:41:49 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/6/2026, 6:30:47 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/3/2026, 1:19:22 PM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 3:41:45 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 3:32:08 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 3:25:08 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 3:23:28 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 3:17:13 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 3:14:02 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 3:07:30 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 3:06:33 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 3:00:09 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 2:57:39 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 1/26/2026, 2:25:35 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 128 | Method unknown | — | 1/26/2026, 1:29:36 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/25/2026, 3:49:49 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/25/2026, 3:33:08 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 1K | Method unknown | — | 1/25/2026, 3:22:58 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/25/2026, 3:14:46 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 1/25/2026, 3:07:29 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 128 | Method unknown | — | 1/25/2026, 3:00:26 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/25/2026, 1:49:45 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 11:31:08 PM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 7:06:39 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 1:54:48 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 9:46:46 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 10:05:45 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 12:50:21 AM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 3:41:11 PM |
| Qwen3-0.6B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:03:04 PM |
| Qwen3-0.6B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/22/2026, 4:33:38 AM |
| Qwen3-0.6B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/21/2026, 12:05:51 AM |
| Qwen3-0.6B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 11:29:48 PM |
| Qwen3-0.6B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:23:44 PM |
| Qwen3-0.6B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 3:28:52 PM |
| Qwen3-1.7B ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 1/23/2026, 3:31:47 AM |
| Qwen3-1.7B ↗ | Text generation | BF16 | vLLM | 1K | Method unknown | — | 1/22/2026, 9:52:43 PM |
| Qwen3-1.7B ↗ | Text generation | BF16 | vLLM | 2K | Method unknown | — | 1/15/2026, 3:23:17 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:11:58 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/6/2026, 6:37:22 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/3/2026, 1:35:40 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 4:29:14 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 4:17:08 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 4:09:27 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 4:08:08 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 4:01:58 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 3:57:42 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 3:50:36 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 3:43:19 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 1K | Method unknown | — | 1/25/2026, 5:14:18 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/25/2026, 5:01:45 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 1/25/2026, 4:49:37 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 128 | Method unknown | — | 1/25/2026, 4:37:27 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 11:47:23 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 7:26:47 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 2:14:41 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 10:01:35 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 10:17:22 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 12:58:00 AM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 3:47:52 PM |
| Qwen3-1.7B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:07:24 PM |
| Qwen3-1.7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/22/2026, 4:35:30 AM |
| Qwen3-1.7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 11:31:38 PM |
| Qwen3-1.7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:25:35 PM |
| Qwen3-1.7B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 3:30:49 PM |
| Qwen3-4B ↗ | Text generation | BF16 | vLLM | 128 | Method unknown | — | 1/23/2026, 3:25:44 AM |
| Qwen3-4B ↗ | Text generation | BF16 | vLLM | 1K | Method unknown | — | 1/22/2026, 9:44:00 PM |
| Qwen3-4B ↗ | Text generation | BF16 | vLLM | 2K | Method unknown | — | 1/15/2026, 3:11:58 PM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 2/6/2026, 12:00:37 PM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 1K | Method unknown | — | 1/25/2026, 6:32:49 AM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 512 | Method unknown | — | 1/25/2026, 6:24:57 AM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 1/25/2026, 6:17:50 AM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 1/25/2026, 6:10:53 AM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 1/23/2026, 2:57:30 AM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 1K | Method unknown | — | 1/22/2026, 8:27:13 PM |
| Qwen3-4B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/15/2026, 2:21:39 PM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 4:47:03 PM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 9:26:36 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 9:19:15 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 9:12:18 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 9:05:17 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 8:36:17 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 7:59:43 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 7:52:31 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 7:46:23 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 7:40:38 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 7:33:24 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 7:09:50 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | Mixed | Method unknown | — | 2/7/2026, 4:15:18 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 128K | Method unknown | — | 2/7/2026, 4:11:10 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 64K | Method unknown | — | 2/7/2026, 4:07:59 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 32K | Method unknown | — | 2/7/2026, 3:50:21 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 16K | Method unknown | — | 2/7/2026, 3:34:14 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 2/7/2026, 2:59:36 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 2/7/2026, 2:46:49 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 2:40:19 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 1K | Method unknown | — | 2/7/2026, 2:34:31 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 2/7/2026, 2:28:53 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/7/2026, 2:22:07 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 128 | Method unknown | — | 2/7/2026, 2:14:44 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/7/2026, 1:29:44 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:24:40 PM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 6:13:24 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 5:59:22 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 5:51:00 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 5:49:07 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 5:43:31 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 5:34:10 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 5:24:50 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 5:16:27 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/23/2026, 12:24:49 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 8:09:57 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 5:14:57 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 3:36:34 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 10:15:35 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 10:52:27 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 1:21:54 AM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 4:07:23 PM |
| Qwen3-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:12:11 PM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:10:28 PM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:43:43 PM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 9:03:58 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 8:50:14 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 8:42:16 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 8:35:09 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/23/2026, 4:00:03 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 8:47:46 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 5:52:24 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 4:14:27 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 10:38:39 AM |
| Qwen3-4B-Instruct-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 8:52:10 PM |
| Qwen3-4B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:39:04 PM |
| Qwen3-4B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:32:56 PM |
| Qwen3-4B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 3:38:19 PM |
| Qwen3-4B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:40:55 PM |
| Qwen3-4B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:34:50 PM |
| Qwen3-4B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 3:40:15 PM |
| Qwen3-4B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 11:42:47 PM |
| Qwen3-4B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:36:44 PM |
| Qwen3-4B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 3:42:08 PM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:02:40 PM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:37:08 PM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 8:07:12 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 7:53:24 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 7:44:54 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 7:36:56 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 128 | Method unknown | — | 1/23/2026, 3:53:56 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 8:34:36 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 5:39:27 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 4:01:14 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 10:26:57 AM |
| Qwen3-4B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 8:40:29 PM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 2/6/2026, 1:02:31 PM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 4K | Method unknown | — | 1/25/2026, 7:35:48 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/25/2026, 7:20:41 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 1K | Method unknown | — | 1/25/2026, 7:10:33 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 512 | Method unknown | — | 1/25/2026, 7:01:23 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 1/25/2026, 6:53:38 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 1/25/2026, 6:45:32 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 1/23/2026, 3:03:39 AM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 1/22/2026, 8:35:44 PM |
| Qwen3-8B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/15/2026, 2:32:20 PM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:18:05 PM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:49:40 PM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 10:01:31 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 9:45:07 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 9:35:21 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 9:27:35 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 9:00:30 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 6:05:20 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 4:27:17 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 10:50:17 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 11:26:08 AM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 7:06:09 PM |
| Qwen3-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:17:29 PM |
| Qwen3-8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:50:20 PM |
| Qwen3-8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:44:28 PM |
| Qwen3-8B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 3:52:22 PM |
| Qwen3-8B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:52:14 PM |
| Qwen3-8B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:46:21 PM |
| Qwen3-8B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 3:54:34 PM |
| Qwen3-8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 11:54:10 PM |
| Qwen3-8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:48:15 PM |
| Qwen3-8B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 3:57:45 PM |
| Qwen3-14B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 2/6/2026, 1:14:50 PM |
| Qwen3-14B-AWQ ↗ | Text generation | AWQ | vLLM | 512 | Method unknown | — | 1/23/2026, 3:09:48 AM |
| Qwen3-14B-AWQ ↗ | Text generation | AWQ | vLLM | 128 | Method unknown | — | 1/22/2026, 8:42:05 PM |
| Qwen3-14B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/15/2026, 2:44:05 PM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:35:46 PM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 1:08:55 PM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 11:58:25 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 11:43:46 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 11:32:13 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 11:24:10 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 9:28:52 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 6:33:27 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 4:55:39 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 11:19:25 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 11:51:54 AM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 7:22:25 PM |
| Qwen3-14B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:23:48 PM |
| Qwen3-14B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:56:08 PM |
| Qwen3-14B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:50:11 PM |
| Qwen3-14B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 4:02:01 PM |
| Qwen3-14B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 11:58:08 PM |
| Qwen3-14B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:52:11 PM |
| Qwen3-14B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 4:07:27 PM |
| Qwen3-14B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/21/2026, 12:00:07 AM |
| Qwen3-14B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:54:11 PM |
| Qwen3-14B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 4:13:41 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:53:56 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 1:27:00 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 2:47:10 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 2:31:20 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 2:18:48 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 2:08:49 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 1/23/2026, 4:14:18 AM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 10:01:13 AM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 5:28:14 AM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 11:33:06 AM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 9:04:30 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 1:35:23 PM |
| Qwen3-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:29:56 PM |
| Qwen3-30B-A3B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/21/2026, 12:02:04 AM |
| Qwen3-30B-A3B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 7:56:10 PM |
| Qwen3-30B-A3B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 4:22:46 PM |
| Qwen3-30B-A3B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 7:58:10 PM |
| Qwen3-30B-A3B-Q8_0 ↗ | Text generation | Q8_0 | vLLM | 1K | Method unknown | — | 1/20/2026, 4:56:37 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 6:03:20 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 1:32:34 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 3:31:53 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 3:15:33 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 3:03:08 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 2:54:22 PM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/23/2026, 3:46:57 AM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 10:16:20 AM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 5:43:44 AM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 10:55:10 AM |
| Qwen3-30B-A3B-Thinking-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 9:16:49 PM |
| Qwen3-32B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 2/6/2026, 1:49:37 PM |
| Qwen3-32B-AWQ ↗ | Text generation | AWQ | vLLM | 256 | Method unknown | — | 1/23/2026, 3:17:48 AM |
| Qwen3-32B-AWQ ↗ | Text generation | AWQ | vLLM | 4K | Method unknown | — | 1/22/2026, 8:48:34 PM |
| Qwen3-32B-AWQ ↗ | Text generation | AWQ | vLLM | 2K | Method unknown | — | 1/15/2026, 2:56:53 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 6:12:14 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 1:38:23 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 4:50:40 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 4:10:15 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 3:48:31 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 3:38:08 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 10:34:40 AM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 6:01:33 AM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 11:46:15 AM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 4:14:36 AM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 7:44:37 PM |
| Qwen3-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/10/2026, 5:40:41 PM |
| Qwen3-32B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 8:00:15 PM |
| Qwen3-32B-Q4_K_M ↗ | Text generation | Q4_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 5:05:10 PM |
| Qwen3-32B-Q5_K_M ↗ | Text generation | Q5_K_M | vLLM | 1K | Method unknown | — | 1/20/2026, 8:02:21 PM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 4:38:34 PM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:18:22 PM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 5:23:39 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 5:11:05 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 5:02:00 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 5:00:40 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 4:54:16 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 4:50:09 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 4:40:56 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 4:31:39 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/23/2026, 12:07:35 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 7:49:56 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 4:54:52 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 3:17:00 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/15/2026, 4:26:39 PM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 10:40:59 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 1:12:54 AM |
| Qwen3-VL-2B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 4:00:25 PM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 4:55:10 PM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:31:10 PM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 7:16:41 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 7:03:19 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 6:54:55 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 6:47:03 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 8:23:18 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 5:28:07 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 3:49:49 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 12:53:18 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 11:02:52 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 1:30:03 AM |
| Qwen3-VL-4B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 4:14:27 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 3/2/2026, 10:19:54 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 3/2/2026, 9:19:18 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 3/2/2026, 9:05:44 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 3/1/2026, 3:52:51 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:26:36 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 12:55:59 PM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 11:06:07 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 10:55:04 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 10:45:38 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 10:38:33 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/23/2026, 4:06:50 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 9:15:47 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 6:20:23 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 4:42:24 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 1:04:48 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 11:38:12 AM |
| Qwen3-VL-8B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 7:13:50 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 5:44:50 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 1:21:01 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 1:26:59 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 1:13:43 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 1:01:39 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 12:52:05 PM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 1/23/2026, 4:21:33 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 9:45:50 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 6:50:34 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 5:12:46 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 1:18:25 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 3:16:16 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/13/2026, 11:57:27 AM |
| Qwen3-VL-30B-A3B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 7:30:30 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/9/2026, 6:31:18 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 256 | Method unknown | — | 2/6/2026, 1:44:06 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 8K | Method unknown | — | 1/26/2026, 5:43:05 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 4K | Method unknown | — | 1/26/2026, 5:20:55 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/26/2026, 5:07:41 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 512 | Method unknown | — | 1/26/2026, 4:57:04 PM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/22/2026, 11:18:25 AM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/20/2026, 1:33:14 AM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 4:56:43 AM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/14/2026, 4:34:55 AM |
| Qwen3-VL-32B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 1/12/2026, 7:53:41 PM |
| Qwen3.5-0.8B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/3/2026, 9:16:20 AM |
| Qwen3.5-2B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/3/2026, 9:23:57 AM |
| Qwen3.5-4B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/3/2026, 9:32:08 AM |
| Qwen3.5-9B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/3/2026, 9:39:06 AM |
| Qwen3.5-27B-FP8 ↗ | Text generation | FP8 | vLLM | 2K | Method unknown | — | 2/27/2026, 12:41:19 PM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | hle-test-self-judge-22ed307-v1-n5-mt16384-t0 | quality-v1 | 20% | 10/17/2026, 2:30:19 AM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | simpleqa-verified-eval-self-judge-0dc97e0-v1-n1000-mt16384-t0 | quality-v1 | 6.4% | 10/17/2026, 1:50:19 AM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | simpleqa-verified-eval-self-judge-0dc97e0-v1-n5-mt16384-t0 | quality-v1 | 0% | 10/15/2026, 9:23:17 PM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | ifbench-test-official-loose-1c40f0c-v1-n300-mt16384-t0 | quality-v1 | 67.67% | 10/15/2026, 8:57:08 PM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | ifbench-test-official-loose-1c40f0c-v1-n3-mt16384-t0 | quality-v1 | 66.67% | 10/15/2026, 9:27:35 AM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | aime2026-test-zero-shot-strict-v4-n30-mt16384-t0 | quality-v1 | 63.33% | 10/15/2026, 9:07:41 AM |
| Qwen3.8-27B-NVFP4 ↗ | Text generation | NVFP4 | vllm | mmlu-pro-test-zero-shot-strict-v4-n3-mt16384-t0 | quality-v1 | 66.67% | 10/15/2026, 6:20:58 AM |
| Step3-VL-10B ↗ | Text generation | Native | vLLM | 2K | Method unknown | — | 3/4/2026, 10:44:11 AM |
Running benchmarks
0Updates automaticallyNo benchmarks are currently reported as running.
Methodology
Average of error-free runs for the selected recorded configuration. Missing methods and fallback runs are not ranked. Performance index = average / fastest matching hardware × 100. Value index = performance per purchase cost / best comparable value × 100. Quality is separate by dataset and protocol. Prices are dated seller listings, not completed transactions or market averages.
Updated 11/5/2026, 9:54:22 PM