MiniMax M2.7 API: pricing, base URL and quick setup

Specifications and pricing

Prices synced with platform billing (USD per 1M tokens).

Quick setup

curl https://a2agent.me/v1/chat/completions \
  -H "content-type: application/json" \
  -H "authorization: Bearer sk-your-a2agent-key" \
  -d '{
    "model": "minimax-m2.7",
    "messages": [{"role": "user", "content": "Reply with OK"}]
  }'

Using it from a coding client

Fine as the edit model in a coding client where each request carries a few files rather than the whole tree. If your sessions routinely run past 200K, M2.5 is the same price with five times the room.

• Run it in Roo Code
• Run it in Cline

How it compares within its family

M2.7 is the later of the two coding-tagged MiniMax models, and the surprise is that it is the shorter one. 200K, where M2.5 runs to 1M. Price is identical across the family, $0.30 per million input tokens and $1.20 output for M2.7, M2.5 and M3 alike, so the newer release costs nothing extra and gives up context. That inverts the usual reasoning about picking the latest model. Choose M2.7 when the task sits comfortably inside 200K and you want the most recent release. Choose M2.5 when the prompt is large and you would rather have the window than the version number. M3 is the third option at 1M, tuned for agent work rather than coding.

Other models on this platform

• MiniMax M2.51M
• MiniMax M31M

Questions