The MiniMax M-series is live
One endpoint, one key, and the MiniMax M-series behind it — pick the model by the job.
We route the M-series through the same endpoint your agent already speaks: set the base URL, paste a key, pick a model per task. Plans have a flat monthly price and hard cap — when you hit it you get a clean 429, not an invoice.
The point of a series is that you don't have to pick one model and live with it. Reach for minimax-m3 when the task needs depth — long agentic runs, gnarly refactors, real reasoning. Run minimax-m2.7 as the daily driver; it's fast, it holds ~205K of context, and it covers most coding and agent work. Drop to minimax-m2.5 when speed and volume matter more than depth — lighter calls, high-frequency loops, quick edits.
Every model in the catalog is available on every plan; the tiers change your monthly allowance, not your model access. Swap between them per request by changing the model id — nothing else in your setup moves.
- minimax-m3The flagship. Long-horizon agentic coding, tool use, and deep reasoning.
- minimax-m2.7The balanced daily driver — ~205K context, fast and strong across coding and agents.
- minimax-m2.5The efficient one. Quick and cheap — ideal for high-volume, lighter calls.