We're thrilled to announce that MiniMax-M2.5 is now live on the NetMind platform with first-to-market API access, free for a limited time! Even before MiniMax officially launches the model!
For your Openclaw agent, Claude Code, or any other AI application, just plug in and build!
The M2 family was designed with agents at its core, supporting multilingual programming, complex tool-calling chains, and long-horizon planning.
M2.5 takes this further with the kind of reliable, fast, and affordable intelligence that makes autonomous AI workflows practical at scale.
Optimized thinking efficiency combined with ~100 TPS output speed delivers approximately 3x faster responses than Opus-class models. For agent loops and interactive coding, that speed compounds fast.
At $0.3/M input tokens, $1.2/M output tokens, $0.06/M prompt caching read tokens, $0.375/M prompt caching write tokens, M2.5 is purpose-built for high-volume, always-on production workloads.
The prompt is as simple as:
E-commerce website for a premium, modular, architectural cat tunnel system. The hero section features a high-definition, cinematic auto-play video of a cat gracefully darting through and playing within a modular tunnel system, capturing fluid movement and texture. Grid-based, spacious, and clean layout utilizing asymmetrical blocks and large-format architectural product photography. Subtle parallax scrolling transitions and an interactive 3D configurator widget demonstrated mid-screen. The style is Scandinavian minimalist meets Wabi-Sabi; refined, peaceful, and quietly luxurious. Muted, earthy tones: warm ivory, soft gray, heather beige, and deep charcoal. High-fidelity photorealism, Figma aesthetics, Architectural Digest quality product visualization.
And here's the stunning result:
Here's a head to head comparison with Claude Opus 4.6:
Start building with M2.5 on NetMind today — the only platform offering first-to-market API access. Limited-time free trial, and many more other models available!
User-Agent: