Speed test matrix
INTELLECT-2 on H200 NVL 141GB
vllm · 0.26.1rc1.dev638+g52be12cfa · BF164 contexts (8K–35K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 35K, 40K
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
- intellect
- 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-primeintellect-vllm/bin/vllm" serve PrimeIntellect/INTELLECT-2 --served-model-name intellect --tensor-parallel-size 1 --reasoning-parser deepseek_r1 --enable-auto-tool-choice --tool-call-parser hermes --max-model-len 40960 --gpu-memory-utilization 0.866040 --host 0.0.0.0 --port 8000 --api-key <redacted> --reasoning-parser deepseek_r1
Prefill
prompt tok/s by context length8K context8,458972 ms TTFT
16K context7,6402,130 ms TTFT
32K context6,6634,862 ms TTFT
35K context6,4985,450 ms TTFT
40K context6,2846,441 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 | 50.363 ms TTFT | 98.287 ms TTFT | 190.8128 ms TTFT | 355.0176 ms TTFT | 625.2258 ms TTFT | Over capacity | Over capacity | Over capacity |
| 16K | 48.989 ms TTFT | 94.2123 ms TTFT | 179.8161 ms TTFT | 317.8187 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | 46.5112 ms TTFT | 87.1119 ms TTFT | 160.4194 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 35K | 46.1121 ms TTFT | 86.0167 ms TTFT | 157.2232 ms TTFT | Over capacity | 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 | 8,458.0 | — | 972 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 7,640.0 | — | 2,130 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 6,663.0 | — | 4,862 ms | — | — | 1 | — | — |
| Prefill | 35K | 1 | Valid | 6,498.0 | — | 5,450 ms | — | — | 1 | — | — |
| Prefill | 40K | 1 | Valid | 6,284.0 | — | 6,441 ms | — | — | 1 | — | — |
| Sustained decode | 8K | 1 | Valid | 50.3 | 50.3 | 63 ms | 19.9 ms | 1.0 | 0 | 1,507 | 30.0 s |
| Sustained decode | 8K | 2 | Valid | 98.2 | 49.1 | 87 ms | 20.4 ms | 2.0 | 0 | 2,946 | 30.0 s |
| Sustained decode | 8K | 4 | Valid | 190.8 | 47.8 | 128 ms | 20.9 ms | 4.0 | 0 | 5,720 | 30.0 s |
| Sustained decode | 8K | 8 | Valid | 355.0 | 44.4 | 176 ms | 22.5 ms | 8.0 | 0 | 10,648 | 30.0 s |
| Sustained decode | 8K | 16 | Valid | 625.2 | 39.2 | 258 ms | 25.5 ms | 16.0 | 0 | 18,752 | 30.0 s |
| Sustained decode | 8K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 8K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 8K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 48.9 | 48.9 | 89 ms | 20.5 ms | 1.0 | 0 | 1,466 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 94.2 | 47.1 | 123 ms | 21.2 ms | 2.0 | 0 | 2,824 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 179.8 | 45.0 | 161 ms | 22.2 ms | 4.0 | 0 | 5,392 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 317.8 | 39.8 | 187 ms | 25.1 ms | 8.0 | 0 | 9,536 | 30.0 s |
| Sustained decode | 16K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 1 | Valid | 46.5 | 46.5 | 112 ms | 21.5 ms | 1.0 | 0 | 1,393 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 87.1 | 43.6 | 119 ms | 22.9 ms | 2.0 | 0 | 2,612 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 160.4 | 40.1 | 194 ms | 24.9 ms | 4.0 | 0 | 4,812 | 30.0 s |
| Sustained decode | 32K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 35K | 1 | Valid | 46.1 | 46.1 | 121 ms | 21.7 ms | 1.0 | 0 | 1,382 | 30.0 s |
| Sustained decode | 35K | 2 | Valid | 86.0 | 43.0 | 167 ms | 23.2 ms | 2.0 | 0 | 2,580 | 30.0 s |
| Sustained decode | 35K | 4 | Valid | 157.2 | 39.4 | 232 ms | 25.4 ms | 4.0 | 0 | 4,716 | 30.0 s |
| Sustained decode | 35K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 35K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 35K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 35K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 35K | 128 | Over capacity | — | — | — | — | — | — | — | — |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.