LLM Inference Benchmark Explorer

DeepSeek-V4.1-Flash on 8× DGX Spark — FP8, vLLM, TP8, speculative decoding k=5 inference benchmark

Open this configuration in the LLM Inference Benchmark Explorer

ModelDeepSeek-V4.1-Flash
Parameters763B
Intelligence Index39.5
Agentic Index—
Device8× DGX Spark
QuantizationFP8
Max C4
Chat Capacity16
Agentic Capacity6
TP8
DP—
PP—
EnginevLLM
Speculative DecodingYes
CTTFT (ms)TPS (tok/s)Status
121335.98PASS
228028.21PASS
437020.10PASS
848813.63FAIL

In OpenZeka's measurement, DeepSeek-V4.1-Flash (763B parameters), served in FP8 format with vLLM and speculative decoding on 8× DGX Spark (TP=8), reached a generation speed of 36 tok/s per request and a time to first token (TTFT) of 213 ms with a single request.

Considering the speed targets and the available KV cache capacity, the estimated capacity is 16 users for chat use and 6 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.

Notes: DSpark k=5, 300K context, Engram in memory (stock vLLM). Model: deepseek-ai/DeepSeek-V4.1-Flash

Intelligence Index and Agentic Index values are published by Artificial Analysis and are reproduced here with attribution.