AI Agent StoreMiniMax · Stable release verified
MiniMax M3.1 Flash · native API
Use text, screenshots and video in coding or multi-step tasks. This is the stable October release, separate from the earlier subscription preview.
Released 2026-10-19 · First reviewed 2026-10-20 · Checked 2026-11-03. The provider dates the stable release October 19. Performance and speed claims in its announcement are vendor evidence, not independent results here.
We still need comparable independent tests. You can compare the confirmed features and limits below, but we cannot yet give this version an Agent Fit or Practical Value rating. Test it on your own tasks before switching; results for another version or thinking setting may not apply.
What you can use today
Token prices measure the small pieces of text the model reads or produces. Capacity is the amount the service accepts in a request; check the limits below for any fallback to another model. A dash means the rate is unverified.
Exact provider API ID: MiniMax-M3.1-Flash. Thinking is always on; requests to turn it off are rejected. Tools can run between thinking steps, but your application must execute them and control permissions. Sampling controls are currently fixed: temperature stays 1 and top_p stays 0.95; supplied values do not change them.
Available thinking settings: low, medium, high, xhigh, max. We do not yet have comparable independent results for these settings. The documented API default is max when effort is omitted. A default is not a recommendation for every task.
Also downloadable as MiniMaxAI/MiniMax-M3.1-Flash. The local template supports low, medium, high, xhigh and max; its recommended sampling settings do not change the hosted API’s fixed controls. Space Bunny was an early version, not an identical tested checkpoint.
The API documents a 1,000,000-token context and up to 524,288 generated tokens. Thinking and the final answer share the output budget.
The direct API charges $0.15 input, $0.015 cached input and $0.60 output per million tokens. These flat rates are separate from M Plan subscription allowances. Priority processing is not supported for this model (the service_tier setting). Downloaded weights use the MiniMax Community License: commercial use requires attribution and a one-time notice; products/services above $20M yearly revenue require prior written authorization. Read the full license for conditions. Allow for retries and any tool fees. Models that think before answering may also charge for that work.
Provider release and specifications ↗
Check availability and plans ↗
More setup and licensing details ↗
More setup and licensing details ↗
More setup and licensing details ↗
Test it before switching
Give each model the same examples, tools and definition of success. Record correct outcomes, response time, retries and total cost. A cheaper token price does not guarantee a cheaper completed task.
Compare caching and long-context costs ↗