AI Hub
All models
LLMOpen weights

MiniMax-M1

MiniMax

Updated May 21, 2026 · Benchmarks via Artificial Analysis, specs & pricing via OpenRouter · Methodology · Report an error

Superseded by MiniMax-M2

Specifications

Type
LLM
Access
Open weights
Released
June 16, 2025
License
Apache 2.0
Parameters
456B (45.9B active)
Context window
1,000,000 tokens
Knowledge cutoff
June 30, 2024
Input
Text
Output
Text
API pricing
$0.40 in · $2.20 out / 1M tokens
Capabilities
Function calling

Benchmarks

Long Context

Our take

First large-scale open-weights hybrid “lightning attention” reasoning model with a native 1M context, designed for efficient long-form test-time compute.

Related research

Links

Compare MiniMax-M1 with

See all MiniMax-M1 alternatives →