What happened
DeepSeek published DeepSeek-V4.1-Flash-1017 on October 17, 2026 as a free, MIT-licensed open model that scored above GPT-6.1 Sol, GPT-6 Astra and Claude Fable 5 on the DeepSWE v1.1 coding benchmark at very low average task cost.
Why it matters
A freely downloadable model matching the top scorers' results at much lower cost gives developers an open alternative for AI coding agents, though whether it holds up across real-world use is not established.
WHO IT HITSIndependent developers and engineering teams building AI coding agents (software that writes and edits code on its own) can now download and run a top-scoring model free under the MIT License, or use it via API, instead of paying for closed models.
Summaries like this, in your inbox every morning.
DeepSeek-V4.1-Flash-1017 is positioned as an update to DeepSeek-V4.1-Flash, with the company claiming both higher processing accuracy and faster task completion. The framing matters because the model's case rests on a combination: a coding-agent benchmark, DeepSWE v1.1, charts average cost per task against score, and DeepSeek-V4.1-Flash-1017 sits above rival models while staying very cheap. It also led several tests against DeepSeek-V4.1-Flash, Kimi K3, GLM-5.3, GPT-5.6 Sol and Claude Opus 5.
The price structure splits peak and off-peak hours. Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday through Friday, and off-peak rates are half price, which gives cost-sensitive teams a reason to schedule work outside those windows.
Availability shapes who can actually use it. At 763 billion parameters, running it on a personal PC is difficult, so practical use means the official API or a third-party API; Ollama and LM Studio have already started hosting it on US servers, which may make it easier for developers to try without managing the model themselves.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Setagaya Ward tested Genspark, an AI agent tool, and cut roughly 2,900 hours of work, according to Sho Miyama…
DeepSeek published the weights of DeepSeek-V4.1-Flash-1017 on Oct 17 under the commercial-use MIT license, cla…
LegalOn Technologies began offering a beta of its "Agent Builder" feature on October 19, which automates a ser…
DeepSeek released DeepSeek-V4.1-Flash-1017 on October 17, a minor update to DeepSeek-V4.1-Flash that it says t…
OpenAI launched ChatGPT Slides, which builds decks from chats, files or outlines and can mimic the design of a…
DeepSeek released Search and Fetch APIs on October 17, 2026, with each endpoint priced at $0.75 per 1,000 call…
