MiniMax M3.1 Flash
A low-cost MiniMax model for agents, coding and computer use, with a 1M-token context window.
Is it actually free? Paid
No free tier has been announced. You pay through the MiniMax API, the M Plan subscription or the MiniMax Code app. MiniMax says it will release the weights later. VentureBeat reports a date of Oct 23, and MiniMax has not announced the license.
Pricing
$0.15 per 1M input tokens, $0.60 per 1M output tokens and $0.015 per 1M cached input tokens.
Best for
High-volume agent and automation work on a tight budget. MiniMax reports a score of 76.2 on BankerToolBench and 73.7% on DeepSWE v1.1. Artificial Analysis gave it an Intelligence Index score of 42 at launch, and its live page now shows 43.
Not great for
Fact-heavy work that nobody checks. VentureBeat reports that Artificial Analysis measured a 93% hallucination rate on AA-Omniscience. When the model gets an answer wrong, it usually guesses instead of declining to answer.