DeepSeek releases V4.1-Flash-1017 with MIT-licensed weights and opens its Search and Fetch APIs
DeepSeek released DeepSeek-V4.1-Flash-1017 on Oct 17, a minor update to V4.1-Flash. DeepSeek says it is live on its website, app and API, and that it finishes tasks faster with shorter outputs than the previous version. The Hugging Face card lists MIT-licensed weights, a 1M token context window and a DeepSWE v1.1 score of 77.0, up from 74.2; the benchmark table is DeepSeek's own. DeepSeek also opened its Search and Fetch APIs to all users at $0.75 per 1,000 calls for each endpoint. Search and Fetch calls through server-side tools in DeepSeek Harness, Claude Code and Codex are free for DeepSeek API users until billing starts on Oct 31.
How you’d use it: Run a coding agent task on V4.1-Flash-1017 through the DeepSeek API.
Sources
- Original source: DeepSeek ↗
- DeepSeek-V4.1-Flash-1017 model card on Hugging Face ↗
- DeepSeek post on the Search and Fetch APIs ↗
- Report on Yahoo Tech ↗