Back to Models

MiniMax M2.1 Highspeed

minimaxminimax/m2.1-highspeed

API guide

Chat completion

Standard OpenAI-compatible chat completion via TheRouter β€” only baseURL and model change from the official OpenAI SDK.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax/m2.1-highspeed",
    "messages": [
      {"role": "system", "content": "You are a helpful coding assistant."},
      {"role": "user", "content": "Write a Rust function to parse a CSV line."}
    ],
    "temperature": 0.3,
    "max_tokens": 4096
  }'

Streaming

For real-time user-facing experiences, streaming benefits most from Highspeed's ~100 tps β€” the faster token delivery directly improves perceived responsiveness.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax/m2.1-highspeed",
    "messages": [{"role": "user", "content": "Write a quick sort function in Python"}],
    "stream": true
  }'

Tool use

M2.1 supports parallel tool calling through its MoE architecture. Highspeed's faster token generation particularly benefits multi-tool agent loops where each tool call followed by a model continuation is a latency-sensitive step.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax/m2.1-highspeed",
    "messages": [
      {"role": "user", "content": "What is the weather in Shanghai and Tokyo?"}
    ],
    "tools": [{
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "Get weather by city",
        "parameters": {
          "type": "object",
          "properties": {
            "city": {"type": "string"}
          }
        }
      }
    }],
    "tool_choice": "auto"
  }'
Fact ledger β€” every claim on this page traces here
sourceURLretrieved
Availabilityplatform.minimax.io β€” Supported Models β†—2026-05-27verified
Output speedplatform.minimax.io β€” Model Overview β†—2026-05-27verified
Architecturegithub.com/MiniMax-AI β€” M2.1 repository β†—2026-05-27verified
Context windowplatform.minimax.io β†—2026-05-27verified
Prompt cachingplatform.minimax.io β€” Model Overview β†—2026-05-27to verify
Licensehuggingface.co/MiniMaxAI β†—2026-05-27to verify
Supported inference backendsgithub.com/MiniMax-AI β†—2026-05-27verified
SWE-bench Verifiedminimax.io β€” official announcement β†—2026-05-27verified
Multi-SWE-benchminimax.io β€” official announcement β†—2026-05-27verified
SWE-bench Multilingualminimax.io β€” official announcement β†—2026-05-27verified
VIBE (aggregate)minimax.io β€” official announcement β†—2026-05-27verified
Toolathlonminimax.io β€” official announcement β†—2026-05-27verified
BrowseCompminimax.io β€” official announcement β†—2026-05-27verified
AIME 2025minimax.io β€” official announcement β†—2026-05-27verified
MMLUminimax.io β€” official announcement β†—2026-05-27to verify
MiniMax releases M2.1 with coding focus and Highspeed variantminimax.io β€” official announcement β†—2026-05-27verified
Does the Highspeed variant produce different output quality vs standard M2.1?AIMLAPI β€” MiniMax HighSpeed guide β†—2026-05-27to verify
Help & contact