What the new model offers
DeepSeek released V4.1-Flash-1017 on October 17, 2026. According to the company, the model scores better on benchmarks, completes tasks faster and produces shorter answers than its predecessor. DeepSeek calls this a "Pareto improvement" in performance and cost, meaning gains on both fronts without a trade-off.
The model is now available on DeepSeek's website, app and API.
Project building and tool support
DeepSeek says the model was improved for end-to-end project development using reinforcement learning with self-feedback. Examples include 3D modeling, a web-based racing game and a Portal 2 recreation built with the Godot engine.
The model is optimized for tools including DeepSeek Harness, Tencent CodeBuddy and WorkBuddy. DeepSeek recommends upgrading to DeepSeek Harness v0.2.1-rc.1 or later for the best experience.
Search APIs open to everyone
DeepSeek opened its Search and Fetch APIs, powered by its in-house search engine, to all users. The company says the infrastructure serves hundreds of millions of requests daily. Each endpoint costs $0.75 per 1,000 calls.
Search and Fetch calls made through server-side tools in DeepSeek Harness, Claude Code and Codex are currently free for DeepSeek API users. Billing starts October 31, 2026, at 16:00 UTC.
Open weights
The model's weights can be downloaded from Hugging Face and ModelScope.
