MiniMax M2.5
MiniMax M2.5 is an AI model by MiniMax that runs in anymize in the EU.
2 min readUpdated
What it is good for
Good for coding and multi-step tasks with tools, at a very low price.
Hosting and operation
MiniMax M2.5 runs in data centres in the EU. It is operated by AWS Bedrock and Inceptron.
For each request anymize picks the operator with the best availability, always within the EU.
Data processing
Data retention at the provider: none. Prompts and requests are not stored by the provider (zero data retention).
Data is not used for training.
With anonymization on, anymize replaces personal data with placeholders before the request reaches the model.
Technical data
| Item | Value |
|---|---|
| Context window | 128,000 tokens |
| Max. output | 16,000 tokens |
| Inputs | Text |
| Capabilities | Tools, Reasoning |
| Class | Swift |
Availability
The model is in the model catalog but not in the default selection. An admin adds it under Governance > Models, in a solo account you do it yourself.
Whether you may use it in a team is set by the “Model access” rule in the Governance Center.