← Models / MiniMax

MiniMax M2.5

230B MoE (10B active, 8 experts per token)

Lightning variant ($0.30/$2.40) available for higher throughput (100 tok/s vs 50 tok/s). Open-source on HuggingFace. Apache 2.0.

chat OpenAI-compatible Function calling 205K context Released 2026-02

Pricing

Type USD / 1M tokens CNY / 1M tokens
Input $0.3 ¥2.1
Output $1.2 ¥8.4

vs Western Models

MiniMax M2.5 vs GPT-5.2 ($1.75/$14)
83% cheaper input 91% cheaper output
MiniMax M2.5 vs Claude Sonnet 4.6 ($3/$15)
90% cheaper input 92% cheaper output
MiniMax M2.5 vs Gemini 2.5 Pro ($1.25/$10)
76% cheaper input 88% cheaper output

IDE Setup

Cursor

base_url: https://api.minimax.chat/v1
model: minimax-m2.5

Cline

base_url: https://api.minimax.chat/v1
model: minimax-m2.5

Strengths

  • + cost-effective
  • + 200K context
  • + Lightning variant for speed

Watch Out For

  • ! content censorship
  • ! less well-known internationally

International Access

Direct access: Yes
Latency from US: 200-400ms
Registration: account
Payment: credit card

Ready to try MiniMax M2.5?

Get your API key from the official platform.

Go to MiniMax →