3.6K views · 33 likes · 2 reposts (at collection)
DeepSeek V4.1 Flash Speeds Rise Up to 1.5x on GB300 NVL72
- Who
- Artificial Analysis
- What
- reported benchmark results up to 1.5x higher tokens per second per agent for DeepSeek V4.1 Flash on GB300 NVL72 with AA-AgentPerf
What happened
Artificial Analysis reports that DeepSeek V4.1 Flash speeds on GB300 NVL72 rose by up to 1.5x tok/s per agent over the past month on its AA-AgentPerf benchmark. It measured 848 tok/s for one agent, up from 596, and about 1.6x the throughput per GPU at about 80 tok/s per agent.
Read next
See the full storyArtificial Analysis Reports Faster DeepSeek V4.1 Flash on GB300 NVL72DeepSeek V4.1 Flash speeds on GB300 NVL72 have increased by up to 1.5x tok/s per agent over the past month on AA-AgentPerf, our benchmark of datacenter systems running realistic agentic coding workloads
This improvement comes from a newer vLLM version, new vLLM kernels for DeepSeek V4.1 on GB300, FlashInfer autotuning, fuller CUDA graph coverage, tuned speculative decoding, and running prefill and decode on the same GPUs in NVIDIA’s updated Dynamo + vLLM setup
➤ Faster speeds: 848 tok/s for a single agent, up from 596 (1.42x)
➤ Higher throughput: ~1.6x the throughput per GPU at ~80 tok/s per agent
The story so far
Artificial Analysis reported that DeepSeek V4.1 Flash reached up to 1.5 times more tokens per second per agent on GB300 NVL72 over the past month, measured with its AA-AgentPerf benchmark of datacenter systems running realistic agentic coding workloads. It recorded 848 tokens per second for one agent, up from 596, and about 80 tokens per second per agent at roughly 1.6 times the throughput per GPU. Artificial Analysis attributed the improvement to a newer vLLM version and new vLLM kernels.
Keep up with AI, one email a day
Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue · 2026-11-07