DeepSeek's V4.1-Flash-1017 beats Claude Opus 5 on several agentic benchmarks, released under MIT license
ChrisGPT · x · 2026-10-17
DeepSeek released V4.1-Flash-1017, which beats Claude Opus 5 on several agentic benchmarks:
- DeepSWE v1.1: 77.0% vs Opus 5's 74.0%
- AutomationBench: 60.5% vs 50.3%
- Apex Bench: 72.1% vs 65.8%
- Agents' Last Exam: 35.4%, also ahead
It's a large step up from the previous V4.1-Flash (e.g. Terminal-Bench 31.2 → 42.3). Opus 5 still leads on HLE, ProgramBench, and NL2Repo-Bench. Model weights are already out under the MIT license.
More from coding & agent
- Hugging Face Hub adds RL Environment category with 153 agent-ready environments — mervenoyann · 2026-10-17
- AI stock-research prompt series: simulate 3 price paths for any ticker over 3-6 months — bigaiguy · 2026-10-17
- AI stock prompt series: scan X for regulatory tailwinds favoring specific companies — bigaiguy · 2026-10-17
- New prototype geoclass pools OpenAI subscriptions to bypass regional inference capacity limits — TheZachMueller · 2026-10-17
- Dumping a huge task into Codex, and it just says "yea, we're doing this son" — wavefnx · 2026-10-17
- Terminal Night 3 report: UNIX philosophy, CLI organization, and AI-era terminal design — 4310sy · 2026-10-17