MiniMax, a Shanghai-based AI startup, said on its WeChat account that it released M3.1 Flash and claimed improvements in coding, agentic tasks, computer use, and cybersecurity.

The company said Space Bunny, which had anonymously processed more than 200 trillion tokens, was an early version of M3.1 Flash.

MiniMax also claimed nearly 3x faster online inference and a 70% cut in usage costs. The source did not specify the comparison baseline.

Public pages show M3.1-Flash-Preview as a coding and research tool and compare it with MiniMax-M3.

An external tracker lists the model’s performance and pricing, but the reviewed snippets did not show benchmark scores or customer pricing.

China’s generative AI rules include filing requirements for qualifying services, but the reviewed sources did not establish the filing status of MiniMax or M3.1 Flash.

🔗 Source: Yicai

Recent MiniMax developments