vLLM boosts DeepSeek-V4.1-Flash throughput up to 5.3x in three weeks
After three weeks of optimization, the vLLM team boosted DeepSeek-V4.1-Flash performance, with agent throughput up 5.3x on the SemiAnalysis AgentX benchmark and TTFT cut nearly 70% at around 100k throughput, aided by SWA bounded replay and DeepSeek's MegaAttention kernels.
2026-10-08 ~ 2026-10-08 · 3 related posts
- Episode 1: vLLM boosts DeepSeek-V4.1-Flash throughput up to 5.3x in three weeks(2026-10-08, 3 posts)
- Episode 2: vLLM Boosts DeepSeek-V4.1-Flash Throughput 5.3x in Three Weeks(2026-10-17, 2 posts)
- Episode 3: DeepSeek V4.1 Flash Shines in Bug Hunt Bench Amid Cheating Allegations(2026-10-17, 11 posts)
- Episode 4: DeepSeek-V4.1-Flash-1017 Released: Community Tests Show Extreme Cost-Performance(2026-10-17, 13 posts)
- Episode 5: DeepSeek V4.1 Flash Tops Open Agentic Coding Leaderboard(2026-10-17, 2 posts)
- Episode 6: Hands-on: DeepSeek V4.1 Flash as Coding Agent Starts Strong, Then Stalls(2026-10-18, 2 posts)
- Episode 7: DeepSeek V4.1 Flash Ranks High Across Benchmarks at Fraction of Cost(2026-10-18, 2 posts)
- DeepSeek-V4.1-Flash on vLLM: 5.3x agentic throughput three weeks after day 0 — vllm_project · 2026-10-08
- vLLM cuts TTFT nearly 70% at ~100K throughput with DeepSeek NVFP4 kernels and fused ops — vllm_project · 2026-10-08
- DeepSeek-V4.1-Flash hits 5x agent throughput on vLLM via SWA bounded replay — vllm_project · 2026-10-08