19 October 2026 · Source: SiliconANGLE
DeepSeek updates its V4.1-Flash model: faster, shorter answers
DeepSeek released an updated version of the V4.1-Flash model it launched in September (V4.1-Flash-1017). The company says the update finishes tasks faster and produces shorter output, though it did not say by how much. The new version is live in DeepSeek's website and mobile apps; API prices are unchanged. In DeepSeek's own testing, the model resolved 77% of tasks on a software engineering test (up from 74.2%). In the same table, Anthropic's Claude Opus 5 still leads on 7 of the 11 published benchmarks. The update follows a nearly five-hour API outage on October 15.
What's in it for you?
If you use the DeepSeek app, you can expect shorter, quicker answers. The numbers come from the company's own testing, so the best approach is to try it side by side with other tools on your own work. Check the privacy terms before entering customer or personal data.
This summary was prepared from the source article with the help of AI. Our editorial policy
More news
22 October 2026 · TechCrunch
Canva launches its first image generation model for all users
22 October 2026 · The Verge
1Password lets AI agents sign in without seeing your passwords
22 October 2026 · Tech & Learning
Tapestry: Free 3D cultural heritage virtual tours
21 October 2026 · The Verge
Anthropic launches Claude Duet agent for small businesses