clef-flash
Run this model
Choose your operating system and enter a model repository. Launch support is checked separately from performance results.
Test conditions
- Task
- Decision · Text → Structured data
- Workload
- Text → Structured data · typed-invoices-v1 · Test duration 33 s
- Batch size
- Batch 64
- Format · Engine
- bf16 · clef-native · PyTorch 2.11.0
- Unit
- decisions/sec
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.
Protocol details
native-forward-v1 · typed-invoices-v1 · decision
GPUs on this test
Only one GPU configuration has matching measurements; a cross-hardware comparison is not yet available.
- RTX PRO 6000 Blackwell Server Edition · 96 GBPyTorch 2.11.0
- Mean throughput
- 15.4decisions/sec
- Observed GPU occupancy
- 21.9GiB
- P95 latency
- 9.66 s
- Reports combined
- 1
1 more GPUs ran this model under other tests
Sortable table
| Hardware | Mean throughput (decisions/sec) | P95 (s) | TTFT (ms) | Observed occupancy (GiB) | GPU power (W) | Throughput / W | VRAM per GPU (GB) | Reports combined |
|---|---|---|---|---|---|---|---|---|
| #1RTX PRO 6000 Blackwell Server Edition · 96 GBPyTorch 2.11.0Runs, parameters and setup ↗ | 15.43 | 9.663 | — | 21.89 | 324.35 | 0.05 | 96 | 1 |
| vGPU-32GB · 32 GBNo matching measurements | — | — | — | — | — | — | 32 | — |
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.
How results are grouped
Results merge successful runs of the same model, precision and workload. CPU thread counts and CUDA wheel builds may differ; actual PyTorch builds are shown. Other recorded protocol settings remain separate.
Updated 11/5/2026, 11:02:27 PM