DeepSeek-V4.1-Flash-1017 Update Tops Open-Weights, Beats Closed Models on Several Agent Benchmarks
teortaxesTex · x · 2026-10-17
DeepSeek released DeepSeek-V4.1-Flash-1017, a minor update with substantial gains across reasoning, coding, agentic tasks, and visual benchmarks while reducing task execution time and output length. Scores improved across the board over the previous Flash: HLE (no tools) 36.8→42.6, Terminal-Bench 4.0 31.2→42.3, ProgramBench 20.3→29.1, Automation-Bench 54.8→60.5, APEX Bench 64.6→72.1.
Third-party comparison tables show the new model beating Kimi K3, GLM-5.3, GPT-6.1 Sol, GPT-6 Astra, and Gemini 4 Argon on DeepSWE (77.0), Automation-Bench, Agents' Last Exam (35.4), and APEX Bench (72.1), reclaiming the strongest open-weights title — though a gap to Claude Opus 5.5 and Gemini 4 Argon remains, and the community awaits a DeepSeek Pro.
More from Models
- Lava lamp test: Gemini 4 nails it for $0.18 in 2 min vs Opus 5.5's $1.27 and 9.5 min — CodeByPoonam · 2026-10-17
- Gemini 4 Argon builds a Minecraft clone in 19 minutes using 6.3M tokens — CodeByPoonam · 2026-10-17
- Scaling01 criticizes Kimi K3 as undertrained: big but starving for tokens — scaling01 · 2026-10-17
- Unreleased Claude model "onigiri-eap-2" surfaces with four reasoning levels — mark_k · 2026-10-17
- OpenAI's internal data: model task time horizons double every 64 days — LeviTurk · 2026-10-17
- DeepSeek-V4.1-Flash-1017 released open-source under MIT license — deepseek_ai · 2026-10-17