MiniMax-M2 API Overview

MiniMax-M2 exposes RESTful endpoints from this deployment so you can run benchmark-grade reasoning and agent workflows without leaving your own stack. The API surface mirrors familiar OpenAI- and Anthropic-style contracts while enforcing this site’s authentication and billing.

• Base URL: https://minimax-m2.com/api/v1
• Authentication: API key issued from the dashboard (Authorization: Bearer <key> for OpenAI-style calls, x-api-key for Anthropic-style calls)
• Models: MiniMax-M2, MiniMax-M2.1, MiniMax-M2.5, MiniMax-M2.7, MiniMax-M3 — see available models for capabilities and selection guidance
• M2.7 implementation guide: MiniMax-M2.7 Guide for model aliases, rollout behavior, and deployment-level pricing controls
• M3 integration guide: MiniMax M3 API Guide for 1M context, multimodal blocks, and protocol-specific thinking controls
• Default model behavior:
  • POST /api/v1/chat/completions: defaults to MiniMax-M2 when model is omitted
  • POST /api/v1/messages: defaults to MiniMax-M2 when model is omitted
  • Hosted web chat UI defaults to MiniMax-M2.5
• Pricing (launch window baseline): Free tier for hosted chat/API evaluation, then $0.50 / million input tokens, $1.50 / million output tokens
  • Note: MiniMax-M2.7 and MiniMax-M3 pricing may be configured per deployment. Use GET /api/v1/models to read effective live rates.

Supported Endpoint Families

• Chat Completions (POST /api/v1/chat/completions): OpenAI-compatible chat API. Supports streaming responses via Server-Sent Events.
• Messages (POST /api/v1/messages): Claude-compatible request body/response format. Currently supports non-stream responses.
• Models (GET /api/v1/models): Returns available models, capabilities, pricing, and channel defaults.

HTTP Environments

Use HTTPS in production. For local testing, the same routes are available on your development server (e.g., http://localhost:3000/api/v1/...).

Authentication Summary

Create API keys in the dashboard and send them with each request:

Authorization: Bearer <api-key>

or for Claude-compatible endpoints:

x-api-key: <api-key>
Anthropic-Version: 2023-06-01

Continue with the following guides:

0. Choose your model — Compare M2, M2.1, M2.5, and M2.7 capabilities and find the best fit
1. Read the M2.7 guide — Understand aliases, fallback behavior, and configurable pricing for production rollout
2. Generate API keys
3. Call Chat Completions
4. Integrate with Claude-style clients