MiniMax M3 Released, a Multimodal and Long-Context Open-Weight Model
MiniMax introduced M3, an open-weight model with frontier-level performance in coding and agentic tasks. It features a 1 million token context window enabled by a new sparse attention architecture called MSA, and is natively multimodal. M3 surpasses GPT-5.5 and Gemini 3.1 Pro on coding benchmarks like SWE-Bench Pro and demonstrates advanced autonomous iteration capabilities, independently optimizing code for a significant speedup in a complex task.
M3 brings advanced multimodal and ultra-long context capabilities to the open-weight community, previously only available in closed-source models. Its significant performance in coding and agentic tasks could accelerate development in these areas.
Learn one new AI thing every day.
Daily Deck sends you seven plain-English cards like this every morning. Free.
Start free