DeepSeek rolls out faster, leaner V4.1-Flash-1017 AI model after service disruptions
DeepSeek has released DeepSeek-V4.1-Flash-1017, a faster, leaner update to its AI model with improved benchmark scores and shorter response times, following recurring service slowdowns and a roughly five-hour outage.
Chinese artificial intelligence start-up DeepSeek has released DeepSeek-V4.1-Flash-1017, a faster, leaner version of its AI model with improved benchmark scores and shorter response times than its predecessor, according to a Bloomberg report carried by Malay Mail.
Released on 17 October, it is available through DeepSeek’s website, app and developer platform.
In results published on Hugging Face, DeepSeek said the new model scored 42.6 on Humanity’s Last Exam without tools — a test of difficult questions spanning mathematics, science and the humanities — up from 36.8 for the previous V4.1-Flash version. Its Terminal-Bench 4.0 score, which assesses coding agents, rose from 31.2 to 42.3, while its DeepSWE v1.1 score climbed from 74.2 to 77.0. These are the company’s own published evaluations, not independent benchmarks, and do not alone establish real-world performance.
The release comes after recurring service slowdowns and an outage of about five hours earlier in the week that left DeepSeek’s servers inaccessible. Some experts attributed the disruption to a shortage of computing resources, as demand for the company’s competitively priced models has strained its services.
DeepSeek’s September launch of the smaller V4.1-Flash model had already pressured rivals: shares of MiniMax Group, Z.ai and Alibaba Group Holding fell on the Hong Kong market after that release, according to Bloomberg.
More on this topic: all Technology stories
