Model release · DeepSeek
DeepSeek-V4.1-Flash-1017
- Model released
- Review updated
The October 17 update of V4.1 Flash. DeepSeek reports HLE (no tools) 42.6, Terminal-Bench 4.0 42.3, DeepSWE v1.1 77.0, ProgramBench 29.1 and NL2Repo-Bench 68.2, with shorter runtimes and outputs. It is tuned for DeepSeek Harness, Tencent CodeBuddy and WorkBuddy; DeepSeek recommends Harness 0.2.1-rc.1 or later. DeepSeek announced it on X at 16:35 Beijing time on October 17, 2026 and released the open weights, saying reinforcement learning with self-feedback improves end-to-end project building.
Availability
API model name deepseek-flash (same calling method and pricing); live on the official DeepSeek web app and mobile app; open weights on Hugging Face and ModelScope; live on OpenRouter with DeepSeek as the first provider, and supported across the Qoder suite; the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp also route here at Flash pricing