DeepSeek ships V4.1 Flash 1017 at the same price and opens its Search and Fetch APIs

DeepSeek has released DeepSeek-V4.1-Flash-1017, an upgraded version of its lightweight model, and switched its website, app and API to it.

The company said on X and in its API docs on Oct 17 that the new model scores higher on benchmarks, finishes tasks faster and writes shorter outputs than the previous V4.1 Flash, calling it a Pareto improvement in performance and cost. The API name deepseek-flash and its pricing are unchanged. Listed scores include 77.0 on DeepSWE v1.1, 42.3 on Terminal-Bench 4.0 and 42.6 on HLE without tools. DeepSeek said reinforcement learning with self-feedback improved end-to-end project building such as game creation and 3D modeling. Weights are on Hugging Face and ModelScope. Its Search and Fetch APIs, backed by an in-house search engine, are now open to everyone at $0.75 per 1,000 calls, and free through server-side tools in DeepSeek Harness, Claude Code and Codex until Oct 31. The announcement drew 3,600 likes in under an hour. TestingCatalog replied "DeepSWE Seek!" and one user wrote "the whale is back." V4.1 Flash launched on Sept 10, and a day earlier the coding agent OpenCode made it free to use.