MiniMax M3
🧠 AI Modelminimax
Multimodal foundation model with 1M-token context, supports text, image, video input.
MiniMax M3 is the latest multimodal foundation model from MiniMax, offering state-of-the-art performance across multiple benchmarks. It accepts text, image, and video inputs and generates text outputs. With a context length of 1,048,576 tokens, it enables handling of extremely long documents and multi-turn conversations. The model supports advanced features like frequency penalty, logit bias, logprobs, and include_reasoning. It achieves high ELO scores in various categories (e.g., 3D: 1331, code: 1310, dataviz: 1296). Pricing is $0.30 per million input tokens and $1.20 per million output tokens.
💡Highlights
- ├─1M-token context window
- ├─Multimodal: text, image, video input
- └─Competitive ELO scores: 1331 (3D), 1310 (code)
🎯For
- ├─AI researchers
- ├─developers building long-context agents
- └─multimodal application developers