DeepSeek Updates V4.1-Flash Model to Improve Performance
The update posts stronger benchmark scores and shorter outputs, while DeepSeek says it improves reasoning, coding and visual tasks.
- On Saturday, Oct 17, Hangzhou-based DeepSeek released its V4.1-Flash-1017 AI model, aiming to provide a faster, leaner alternative while intensifying competition with Silicon Valley rivals OpenAI and Anthropic.
- The company positioned V4.1-Flash-1017 as a minor update to its September release of the V4.1-Flash. Aggressive pricing previously caused shares of Alibaba Group Holding and MiniMax Group to plummet on the Hong Kong stock exchange.
- Benchmark testing showed "substantial improvements across reasoning, coding, agentic tasks and visual benchmarks," the Hangzhou-based start-up said. On Humanity's Last Exam , the model improved its score to 42.6 from the prior version's 39.1.
- The model is now live across the company's app and developer platform, offering "stronger benchmark performance, faster task completion, and shorter outputs than the previous version," according to posts on X.
- Service disruptions and slowdowns have increased in recent months for DeepSeek, whose aggressively priced models strain its infrastructure. Earlier this week, the company suffered a five-hour outage that some experts attributed to insufficient computing resources.
39 Articles
39 Articles
DeepSeek Releases V4.1-Flash-1017 With Higher Agent and Coding Benchmark Scores
DeepSeek released V4.1-Flash-1017 on October 17, a minor update to its September V4.1-Flash model. In DeepSeek's own tests, DeepSWE rose to 77.0 and Terminal-Bench 4.0 to 42.3, with faster task completion and shorter outputs. The weights are on Hugging Face under the MIT license.
DeepSeek updates V4.1-Flash model to improve performance
According to DeepSeek, the updated model has 'substantial improvements' across reasoning, coding, agentic tasks and visual benchmarks. Read more: DeepSeek updates V4.1-Flash model to improve performance
Version V4.1-Flash-1017 is now available on the web, app, and developer platform. Asian company says it improves programming, reasoning, and tasks with agents, while facing growing competition and availability issues.
Beijing, Oct 19 (EFE).- Chinese technology DeepSeek announced an update of its artificial intelligence (IA) model V4.1 Flash, with improvements in programming and autonomous execution of tasks, and the opening of its own search service for developers. The new version, called DeepSeek-V4.1-Flash-1017 and presented this weekend, is now available in [...] The DeepSeek entry improves its AI for programming and opens its own developer search engine w…
DeepSeek releases V4.1 Flash-1017 with stronger coding and agent capabilities · TechNode
DeepSeek released V4.1 Flash-1017 on October 17, updating its Flash model with improvements in reasoning, coding, agent tasks and visual understanding. The model weights are available on Hugging Face under the MIT license. The company said the update reduces task execution time and output length, allowing users to complete work faster and at lower cost. […]
On the 17th, DeepSeek, a Chinese artificial intelligence (AI) startup, unveiled 'V4.1 Flash-1017', an AI model with enhanced inference, coding, and AI agent functions. The latest model updates (improvements) the performance of the “V4.2-Flash,” which was released last month. DeepSeek's latest model “completes tasks faster than previous versions, and answers are shorter,” and “inference, coding, and AI agent work
Coverage Details
Bias Distribution
- 50% of the sources are Center, 50% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium



















