LLM Inference Benchmark Explorer
DeepSeek-V4.1-Flash-1017 on DGX B300 — FP8, vLLM, TP2, speculative decoding k=3 inference benchmark
Open this configuration in the LLM Inference Benchmark Explorer
| Model | DeepSeek-V4.1-Flash-1017 |
|---|---|
| Parameters | 763B |
| Intelligence Index | — |
| Agentic Index | — |
| Device | DGX B300 |
| Quantization | FP8 |
| Max C | 256+ |
| Chat Capacity | 1024+ |
| Agentic Capacity | 384+ |
| TP | 2 |
| DP | — |
| PP | — |
| Engine | vLLM |
| Speculative Decoding | Yes |
| C | TTFT (ms) | TPS (tok/s) | Status |
|---|---|---|---|
| 1 | 37 | 402.97 | PASS |
| 2 | 47 | 326.75 | PASS |
| 4 | 55 | 280.68 | PASS |
| 8 | 68 | 223.49 | PASS |
| 16 | 87 | 161.57 | PASS |
| 32 | 110 | 132.71 | PASS |
| 64 | 115 | 101.10 | PASS |
| 128 | 189 | 69.49 | PASS |
| 256 | 490 | 45.68 | PASS |
In OpenZeka's measurement, DeepSeek-V4.1-Flash-1017 (763B parameters), served in FP8 format with vLLM and speculative decoding on DGX B300 (TP=2), reached a generation speed of 403 tok/s per request and a time to first token (TTFT) of 37 ms with a single request.
Considering the speed targets alone (the KV cache limit was not calculated for this configuration), the estimated capacity is at least 1,024 users for chat use and at least 384 for agentic use. The realistic capacity will likely fall between these two values. In scenarios dominated by coding, tool use, long workflows and multi-agent use, capacity approaches the agentic estimate; where shorter interactions, standard conversations and lighter tasks dominate, it approaches the chat estimate.
“At least” means the configuration met the speed targets even at the highest load tested, so the real capacity may be higher.
Notes: DSpark k=3, 1M context. Model: deepseek-ai/DeepSeek-V4.1-Flash-1017
Intelligence Index and Agentic Index values are published by Artificial Analysis and are reproduced here with attribution.
In OpenZeka's measurement, DeepSeek-V4.1-Flash-1017 (763B parameters), served in FP8 format with vLLM and speculative decoding on DGX B300 (TP=2), reached a generation speed of 403 tok/s per request and a time to first token (TTFT) of 37 ms with a single request.
Considering the speed targets alone (the KV cache limit was not calculated for this configuration), the estimated capacity is at least 1,024 users for chat use and at least 384 for agentic use. The realistic capacity will likely fall between these two values. In scenarios dominated by coding, tool use, long workflows and multi-agent use, capacity approaches the agentic estimate; where shorter interactions, standard conversations and lighter tasks dominate, it approaches the chat estimate.
“At least” means the configuration met the speed targets even at the highest load tested, so the real capacity may be higher.