MiniMax M3: Open-weight model with a million-token context challenges proprietary leaders

MiniMax released M3, an open-weight multimodal model with a one-million-token context window.

The company says a new sparse-attention design makes that long context practical without extreme compute cost.

M3 reportedly performs strongly on coding, terminal, web search, and autonomous multi-hour tasks, with results near leading proprietary systems on several benchmarks.

The model highlights how open-weight AI is closing the gap on long-context reasoning and multimodal capability while lowering the cost of very large contexts.