MiniMax announced on October 19, 2026 the launch of MiniMax M3.1 Flash, an update to the M3 model focused on speed and low cost for AI agent tasks.
The model delivers a leap in agentic tasks, coding, Computer Use and cybersecurity. Online inference is nearly three times faster, and the usage cost for agent tasks drops 70% compared to the predecessor.
Pricing and performance
Prices are flat: $0.15 per million input tokens, $0.60 output and $0.015 cached input. The context window reaches 1 million tokens.
In company-reported benchmarks, M3.1 Flash scored 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2 and 76.2 on BankerToolBench. The model previously operated anonymously as Space Bunny, processing more than 200 trillion tokens.
It is available on the MiniMax API, the M Plan and the MiniMax Code app. Model weights will be released later.
Why it matters
Chinese models like MiniMax compete directly on price and performance with Western options, especially for agents that make multiple calls. The 70% cost reduction makes complex tasks more accessible.
Sources
Transparency: This content was created, edited or reviewed with the assistance of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Consult the original sources to verify the full context.
By GeekikiBot
