How DeepSeek launched DeepSeek-V4.1-Flash-1017 on X

DeepSeek-V4.1-Flash-1017 is an update to DeepSeek's Flash model for teams running coding agents, terminal work, and long-context multimodal tasks. It shipped on October 17, 2026 to the company's website, app, and API, with weights on Hugging Face and ModelScope under the MIT license. DeepSeek, the Hangzhou AI company founded in 2023 by Liang Wenfeng, says the checkpoint scores higher than the previous V4.1-Flash on reasoning, coding, agent, and visual tests while finishing tasks faster and producing shorter outputs. DeepSeek's model card shows gains on all 11 benchmarks it reports. Terminal-Bench 4.0 rises from 31.2 to 42.3, ProgramBench from 20.3 to 29.1, and DeepSWE v1.1 from 74.2 to 77.0. In that table, the DeepSWE score sits above Opus 5 at 74.0 and GPT-5.6 Sol at 73.0. HLE without tools moves from 36.8 to 42.6. The release lists a 1 million token context window and 763 billion parameters. DeepSeek says the checkpoint uses reinforcement learning with self-feedback for end-to-end project work, including games and 3D scenes, and is tuned for DeepSeek Harness, Tencent CodeBuddy, and WorkBuddy. Harness users are told to upgrade to v0.2.1-rc.1 or later. The same announcement opens Search and Fetch APIs on DeepSeek's in-house search engine, which the company says already serves hundreds of millions of requests a day. Each endpoint costs $0.75 per 1,000 calls. Calls made through server-side tools in DeepSeek Harness, Claude Code, and Codex stay free for DeepSeek API users until billing begins on October 31, 2026 at 16:00 UTC.