重點導讀
DeepSeek V4.1 在 GB300 NVL72 上的 Flash 速度在 AA-AgentPerf 基準中提升至 1.5 倍。
- 單代理速度：848 tok/s，較上月的 596 tok/s 上升 1.42 倍。
AI 伴隨問答
針對這篇報導有疑問？可直接點選快捷問題或向 AI 追問背景脈絡與技術細節：
Artificial Analysis@ArtificialAnlys
·2 小時前
DeepSeek 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
推文來源 · X (Twitter)在 X 參與討論
來源： Artificial Analysis (x.com)
在 X 查看推文