88Z AI, 20 October — MiniMax has released the new generation model MiniMax M3.1 Flash, achieving generational improvements in agentic tasks, programming, and cybersecurity. It offers approximately three times faster inference speed and a seventy percent reduction in comprehensive costs, with top domestic rankings in multiple benchmarks.
At a glance
- Inference speed increased by approximately three times with a seventy percent reduction in comprehensive agentic costs
- Ranked first among domestic models in four benchmarks including DeepSWE
The update
MiniMax has officially launched the new generation Flash model, MiniMax M3.1 Flash. This model achieves generational performance improvements in agentic tasks, programming capabilities, computer use, and cybersecurity. Its inference speed is approximately three times faster, while significantly reducing comprehensive agentic costs by seventy percent. In multiple authoritative benchmark tests, M3.1 Flash has performed excellently. It scored seventy-three point seven percent in DeepSWE, sixty-one point seven in OSWorld V2, and twenty-nine point six in ALE, ranking first among domestic models in these metrics. Additionally, the model achieved an eighty-two percent score in CyberGym. Previously, the anonymous Space Bunny had processed over two hundred trillion tokens, laying the foundation for this release.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- MiniMax 稀宇科技Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.