MiniMax-M3 hits OpenRouter with 50% discount

MiniMax-M3 showcases the rapid acceleration of open-weight models, offering extreme long-context capabilities and multimodal features that were previously exclusive to proprietary models, but its ultimate impact will depend on real-world inference costs and reliability under complex agentic workloads.

• –The 1M-token context window provides a substantial competitive advantage for enterprise applications handling large documents or video files.
• –Native video and image support makes it highly versatile for multimodal AI pipelines.
• –OpenRouter's 50% discount strategy is an excellent tactic to incentivize immediate developer experimentation.
• –Leveraging MiniMax Sparse Attention (MSA) is key to making large-context inference both fast and cost-effective.