DeepSeek Releases AI Model Update, Opens Search APIs to All Users

3-line summary

• Chinese AI startup DeepSeek released DeepSeek-V4.1-Flash-1017, available on its website, mobile app, and API, the company announced via its official X account Saturday; the update improves benchmark performance, speeds up task completion, and shortens output length, which the company described as a 'Pareto improvement' in capability and efficiency without added cost.
• The model is optimized for tools including DeepSeek Harness, Tencent CodeBuddy, and WorkBuddy, with self-feedback-based reinforcement learning the company said improves end-to-end project development such as game creation and 3D model generation; benchmark scores rose on DeepSWE v1.1 (74.2 to 77.0) and Automation-Bench V1.0.6 (54.8 to 60.5).
• DeepSeek also opened its Search and Fetch APIs, built on its own search engine, to all users at $0.75 per 1,000 calls per endpoint; the report did not address market or competitor impact.

This summary is our own processing of the source article; see the source link below for full context.