MiniMax
MiniMax's text, video, speech and music models, served through its open platform.
At a glance
Current release
From the release row MiniMax-M3. Base tier (<=512k tokens); above 512k tokens $0.60 input / $2.40 output; page shows a 'Permanent 50% off' label
Family record: Official documentation · 5 Sep 2026
Built on it
1 product in this catalogue names this family on their own pages.
Releases
-
MiniMax-Text-01
456B-parameter mixture-of-experts language model (45.9B active) using Lightning Attention with a 4 million token context, released with MiniMax-VL-01.
-
MiniMax-M1
Open-weight hybrid-attention reasoning model, 456B parameters (45.9B active), 1 million token context and up to 80,000 tokens of reasoning output; 40k and 80k variants.
-
MiniMax-M2
Model built for agentic and coding work, with open weights on MiniMax's Hugging Face organisation; M2.1 followed on 22 Dec 2025.
-
MiniMax-M2.5
M2 line update (with M2.5-highspeed) aimed at programming, tool calling, search and office work; only the month was stated. M2.1 (Dec. 22, 2025) preceded it.
-
MiniMax-M2.7
M2.7 and a M2.7-highspeed variant, described as emphasizing recursive self-improvement.
-
MiniMax-M3
Natively multimodal model, about 428B parameters (23B active) with MiniMax Sparse Attention and up to 1M token context; API on release, weights on Hugging Face later.