started · updated
DeepSeek releases V4.1-Flash-1017 AI model with improved efficiency
DeepSeek has released an updated version of its AI model, DeepSeek-V4.1-Flash-1017, on October 17, 2026. The company describes the update as a “Pareto improvement” in performance and cost, offering faster task completion, shorter outputs, and improved benchmark scores compared to the previous version.
In software engineering benchmarks, the model achieved a score of 77.0 on the DeepSWE v1.1 test, surpassing previous versions and competing closely with models from Anthropic and OpenAI. The update also shows improvements in reasoning, coding, agentic tasks, and visual benchmarks. The model weights have been released under the MIT license on Hugging Face and ModelScope.
Additionally, DeepSeek has opened its Search and Fetch APIs to all users. These tools, powered by an in-house search engine, allow developers to retrieve live web content and Markdown-formatted pages. The Search and Fetch API endpoints are priced at $0.75 per 1,000 successful calls. The model is currently available via the DeepSeek website, mobile application, and developer API platform.
Entities
Anthropic · DeepSeek · Hugging Face · ModelScope · Nvidia · OpenAI · Tencent
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 21 SOURCES] The updated model provides faster task completion and shorter outputs compared to the previous version. techfocus24.com · www.brasilemfolhas.com.br · www.bloomberght.com · www.techi.com · the-decoder.de · +15 more
- [● 11 SOURCES] The Search and Fetch APIs are now open to all users. techfocus24.com · ppc.land · www.tokenpost.kr · savedelete.com · tbreak.com · +6 more
- [● 9 SOURCES] The model weights for V4.1-Flash-1017 are released under the MIT license. www.techi.com · the-decoder.de · www.tokenpost.kr · savedelete.com · tbreak.com · +3 more
- [● 7 SOURCES] The model uses reinforcement learning with self-feedback (RLSF) to improve project development. techfocus24.com · the-decoder.de · econotimes.com · ppc.land · www.trendingtopics.eu · +2 more
- [● 8 SOURCES] The V4.1-Flash-1017 update delivers a Pareto improvement in performance and cost. techfocus24.com · econotimes.com · savedelete.com · the-decoder.de · nordictimes.com · +3 more
- [● 8 SOURCES] API pricing for the Flash model remains unchanged following the update. www.techi.com · exame.com · www.tokenpost.kr · tbreak.com · www.blocktempo.com · +3 more
- [● 4 SOURCES] The model is optimized for tools including DeepSeek Harness, Tencent CodeBuddy, and WorkBuddy. techfocus24.com · econotimes.com · the-decoder.de · www.trendingtopics.eu
- [● 7 SOURCES] The Search and Fetch API endpoints cost $0.75 per 1,000 calls. techfocus24.com · ppc.land · www.tokenpost.kr · savedelete.com · tbreak.com · +2 more
- [● 14 SOURCES] DeepSeek released the V4.1-Flash-1017 model on October 17, 2026. techfocus24.com · www.tokenpost.kr · nordictimes.com · savedelete.com · tbreak.com · +9 more
- [● 6 SOURCES] Each Search or Fetch API endpoint costs $0.75 per 1,000 calls. techfocus24.com · www.tokenpost.kr · savedelete.com · tbreak.com · cryptobriefing.com · +1 more