Speed test matrix
Olmo-3.1-7B-RL-Zero-Code on H200 NVL 141GB
vllm · 0.26.1rc1.dev638+g52be12cfa · BF164 contexts (8K–59K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 59K, 64K
Hardware
- Type
- Discrete GPU
- GPU
- H200 NVL
- VRAM
- 141 GB
- OS
- linux
Engine
- Engine
- vllm
- Version
- 0.26.1rc1.dev638+g52be12cfa
- Quantization
- BF16
- Backend
- cuda
- Repository
- github.com/vllm-project/vllm
- Build
- cuda-13.2
- Commit
52be12cfac0c
Run
- Served model
- olmo
- Revision
- main
- Benchmark tool
- llm-inference-bench v0.4.29
- Decode test
- Fixed duration
- Prefill mode
- Measured during decode warm-up
- Time per measurement
- 30 s
- Max output
- 4,096 tokens
Launch command
"$HOME/env_h200-allenai-vllm/bin/vllm" serve allenai/Olmo-3.1-7B-RL-Zero-Code --served-model-name olmo --trust-remote-code --tensor-parallel-size 1 --max-model-len 65536 --gpu-memory-utilization 0.869211 --host 0.0.0.0 --port 8000 --api-key <redacted>
Prefill
prompt tok/s by context length8K context30.2K178 ms TTFT
16K context29.2K363 ms TTFT
32K context29.5K712 ms TTFT
59K context28.2K1,368 ms TTFT
64K context28.2K1,485 ms TTFT
Sustained decode
total tok/s · rows are context length, columns are simultaneous requests| Context | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128 |
|---|---|---|---|---|---|---|---|---|
| 8K | 166.851 ms TTFT | 305.754 ms TTFT | 516.278 ms TTFT | 778.7150 ms TTFT | 1,036.0175 ms TTFT | 1,257.9325 ms TTFT | 1,423.2553 ms TTFT | Over capacity |
| 16K | 162.371 ms TTFT | 294.492 ms TTFT | 487.487 ms TTFT | 714.0179 ms TTFT | 916.4263 ms TTFT | 1,074.0359 ms TTFT | Over capacity | Over capacity |
| 32K | 154.396 ms TTFT | 274.585 ms TTFT | 442.3178 ms TTFT | 612.0192 ms TTFT | 753.1327 ms TTFT | Over capacity | Over capacity | Over capacity |
| 59K | 142.3145 ms TTFT | 245.5148 ms TTFT | 380.5251 ms TTFT | 492.6334 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.
TTFT is measured after the benchmark warms the server's prompt cache, so it mostly excludes prefill.
All measurements
Latencies are medians. Total tok/s is output across all simultaneous requests.| Phase | Context | Requests at once | Status | Total tok/s | Per request tok/s | TTFT | Time between tokens | Average active requests | Completed requests | Output tokens | Measured for |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Prefill | 8K | 1 | Valid | 30.2K | — | 178 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 29.2K | — | 363 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 29.5K | — | 712 ms | — | — | 1 | — | — |
| Prefill | 59K | 1 | Valid | 28.2K | — | 1,368 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 28.2K | — | 1,485 ms | — | — | 1 | — | — |
| Sustained decode | 8K | 1 | Valid | 166.8 | 167.7 | 51 ms | 6.0 ms | 1.0 | 1 | 5,005 | 30.0 s |
| Sustained decode | 8K | 2 | Valid | 305.7 | 154.2 | 54 ms | 6.5 ms | 2.0 | 2 | 9,169 | 30.0 s |
| Sustained decode | 8K | 4 | Valid | 516.2 | 133.5 | 78 ms | 7.5 ms | 4.0 | 4 | 15,487 | 30.0 s |
| Sustained decode | 8K | 8 | Valid | 778.7 | 98.5 | 150 ms | 10.2 ms | 8.0 | 0 | 23,361 | 30.0 s |
| Sustained decode | 8K | 16 | Valid | 1,036.0 | 65.5 | 175 ms | 15.3 ms | 16.0 | 0 | 31,072 | 30.0 s |
| Sustained decode | 8K | 32 | Valid | 1,257.9 | 39.7 | 325 ms | 25.2 ms | 32.0 | 0 | 37,728 | 30.0 s |
| Sustained decode | 8K | 64 | Valid | 1,423.2 | 22.4 | 553 ms | 44.7 ms | 64.0 | 0 | 42,688 | 30.0 s |
| Sustained decode | 8K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 162.3 | 163.3 | 71 ms | 6.1 ms | 1.0 | 1 | 4,869 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 294.4 | 148.8 | 92 ms | 6.7 ms | 2.0 | 2 | 8,832 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 487.4 | 124.9 | 87 ms | 8.0 ms | 4.0 | 4 | 14,627 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 714.0 | 90.0 | 179 ms | 11.1 ms | 8.0 | 0 | 21,432 | 30.0 s |
| Sustained decode | 16K | 16 | Valid | 916.4 | 57.7 | 263 ms | 17.3 ms | 16.0 | 0 | 27,504 | 30.0 s |
| Sustained decode | 16K | 32 | Valid | 1,074.0 | 33.8 | 359 ms | 29.6 ms | 32.0 | 0 | 32,224 | 30.0 s |
| Sustained decode | 16K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 1 | Valid | 154.3 | 155.5 | 96 ms | 6.4 ms | 1.0 | 1 | 4,630 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 274.5 | 138.9 | 85 ms | 7.2 ms | 2.0 | 2 | 8,234 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 442.3 | 111.1 | 178 ms | 9.0 ms | 4.0 | 0 | 13,268 | 30.0 s |
| Sustained decode | 32K | 8 | Valid | 612.0 | 77.0 | 192 ms | 13.0 ms | 8.0 | 0 | 18,360 | 30.0 s |
| Sustained decode | 32K | 16 | Valid | 753.1 | 47.3 | 327 ms | 21.1 ms | 16.0 | 0 | 22,592 | 30.0 s |
| Sustained decode | 32K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 59K | 1 | Valid | 142.3 | 144.0 | 145 ms | 6.9 ms | 1.0 | 1 | 4,268 | 30.0 s |
| Sustained decode | 59K | 2 | Valid | 245.5 | 126.8 | 148 ms | 7.9 ms | 2.0 | 2 | 7,364 | 30.0 s |
| Sustained decode | 59K | 4 | Valid | 380.5 | 95.6 | 251 ms | 10.5 ms | 4.0 | 0 | 11,412 | 30.0 s |
| Sustained decode | 59K | 8 | Valid | 492.6 | 61.9 | 334 ms | 16.2 ms | 8.0 | 0 | 14,776 | 30.0 s |
| Sustained decode | 59K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 59K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 59K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 59K | 128 | Over capacity | — | — | — | — | — | — | — | — |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.