DeepSeek V4.1-Flash-1017 Released: Major R&D Improvements and New Search API
Version Overview
On October 17, 2026, DeepSeek officially announced the update of V4.1 Flash to V4.1-Flash-1017. This release is simultaneously available on the web interface, mobile app, and API, with the new model achieving significant improvements across all benchmark tests.
Key Updates
- Significantly Enhanced R&D Capabilities: The new model performs better in code generation, project analysis, and complex task handling, enabling more efficient completion of development-related work.
- Lower Costs: While maintaining high performance, inference costs have been further optimized, providing users with a more cost-effective option.
- Faster Response Speed: Task completion speed has noticeably improved, delivering a smoother user experience.
- First-Ever Search and Fetch API: New web search and content fetching endpoints are now available at 5 yuan per 1,000 calls, officially open to the public.
API Pricing
The new endpoints are priced as follows:
- Search endpoint: 5 yuan per 1,000 calls
- Fetch endpoint: 5 yuan per 1,000 calls
The API pricing for V4.1-Flash-1017 itself remains unchanged, continuing the Flash series' excellent value proposition.
Upgrade Recommendations
All DeepSeek users are encouraged to upgrade to the latest version to experience the following improvements:
- Developers gain stronger code assistance and project analysis capabilities
- Regular users enjoy faster response speeds and lower costs
- Developers can build more powerful AI applications using the new Search and Fetch APIs
Summary
The release of DeepSeek V4.1-Flash-1017 marks an important advancement in the series' practicality and functionality. The new search APIs open up new possibilities for developers, while the model's inherent capability improvements benefit all users. As a leading Chinese large language model, DeepSeek continues to iterate rapidly, closing the gap with top international models.