vGPU-32GB · 32 GB
Kokoro-82M
Open a report for complete measurements.
- Task
- Speech synthesis · Text → Audio
- Workload
- Text → Audio · kokoro-english-v1 · Test duration 6 s
- Batch size
- Batch 2
- Format · Engine
- fp32 · kokoro · PyTorch 2.13.0+cu130
- Unit
- audio-sec/sec
- Mean throughput
- 79.8audio-sec/sec
- Observed GPU occupancy
- 1 GiB
- P95 latency
- 100 ms
- Reports combined
- 1
Launch this model
This engine does not yet have an executable serving recipe.
Individual tests
Text → Audio · Speech synthesis · fp32 · kokoro · tts · audio-sec/sec
Sortable table
| Test date | Throughput | Concurrency / batch | P95 (s) | TTFT (ms) | Observed occupancy (GiB) | GPU power (W) | Decode (tok/s) | Uncached prefill (tok/s) | Parameters | View report |
|---|---|---|---|---|---|---|---|---|---|---|
| 10/8/2026, 3:30:29 AM | 79.84 audio-sec/sec | 2 | 0.1 | — | 1.03 | 100.82 | — | Uncached prefill unavailableView report | Parameters ↗ | d90a5c29 ↗ |
Recorded test parameters
Configuration 1 · Parameters
- model
- hexgrad/Kokoro-82M
- engine
- kokoro
- quant
- fp32
- method
- audio-native-v1
- task
- tts
- dataset
- kokoro-english-v1
- context
- tts
- input_tokens
- 0
- output_tokens
- 0
- duration
- 6
- concurrency
- 2
- unit
- audio-sec/sec
- software
- { "torch": "2.13.0+cu130", "transformers": "5.19.0", "sentence-transformers": "6.1.0", "ctranslate2": "4.8.2", "faster-whisper": "1.2.1", "kokoro": "0.9.4", "OMP_NUM_THREADS": "4", "MKL_NUM_THREADS": "4", "OPENBLAS_NUM_THREADS": "1", "NUMEXPR_NUM_THREADS": "4" }
- engine_provenance
- —
- kernel_settings
- —
- precision
- —
- resolution
- —
- performance_task
- —
- performance_scope
- —
- batch_size
- —
- dimensions
- —
- weight_quantization
- —
- max_sequence_length
- —
- model_scope
- —
- performance_dataset
- —
- workload_mode
- —
- graph_mode
- —
- model_revision
- —
- client_version
- python-native-bench-1.1.3