The latest AI news and updates of MiniMax — AI company building multimodal models for text, voice, image, and video generation. Covering MiniMax's latest product updates and launches from the past 90 days.
MiniMax15h agoMiniMax released M3 open weights, a multimodal model featuring a 1-million-token context window and a new sparse attention architecture. Modular is a day-zero launch partner, providing kernel-optimized inference on its cloud. The model, which uses approximately 428 billion total parameters and 23 billion activated parameters, is designed for frontier coding and long-horizon agentic tasks.
MiniMaxJun 10MiniMax has made its multimodal AI models available on the Respan AI Gateway. This provides developers with a unified endpoint to manage, route, and deploy MiniMax's text, speech, image, and video generation capabilities in production. The gateway offers features like failover, caching, and spend controls to streamline AI product development.
MiniMaxJun 4MiniMax announced that its M3 model is joining the NVIDIA and Microsoft local LLM lineup, with weights releasing to the community within 10 days. The move brings high-capacity multimodal reasoning and coding capabilities directly to local hardware.
MiniMaxJun 4MiniMax announced a partnership with Mem0 to integrate a persistent memory layer into its new M3 multimodal model. By combining a 1-million-token context window with long-term data retention, the integration allows developers to build autonomous agents that remember user preferences across multiple sessions.
MiniMaxJun 3MiniMax revealed technical highlights for its M3 model, featuring a Sparse Attention architecture that maintains uncompressed data for its 1-million-token context window. The update reduces attention kernel overhead from 30% to 5% of per-decode wall-clock time and introduces vision-coding capabilities where the model self-evaluates its own rendered UI.
MiniMaxJun 3MiniMax M3 is now available on SiliconFlow, bringing frontier-grade agentic coding and a million-token context window to the open-weight ecosystem. The launch includes a week-long introductory discount, making high-capacity multimodal reasoning significantly more accessible for developers.
MiniMaxJun 2Cloudflare has integrated the MiniMax M3 foundation model into its AI Gateway platform. The update provides developers with a high-context, multimodal model specialized for autonomous coding tasks directly within their existing infrastructure.
MiniMaxMay 20MiniMax integrated its Speech 2.8 Turbo model into Together AI, adding over 600 expressive voices to the platform's catalogue. This expansion provides developers with high-fidelity, low-latency audio synthesis specifically optimized for building autonomous voice agents on dedicated infrastructure.
MiniMaxApr 17MiniMax partnered with Nous Research to launch MaxHermes, a managed cloud version of the open-source Hermes Agent powered by the MiniMax M2.7 model. The platform enables 10-second deployment of agents that autonomously learn new skills and integrate directly into messaging apps like Slack and Discord.
MiniMaxApr 15MiniMax open-sourced three Music Skills that allow AI agents to generate full tracks, sing in character, and curate local music libraries. By moving music generation from a standalone tool to a native agent capability, developers can now build multimodal agents that use audio as a functional output.
MiniMaxApr 13MiniMax released MMX-CLI, a command-line interface that provides AI agents with native capabilities for image, video, music, and voice generation. By treating these multimodal outputs as terminal commands, agents can now autonomously create and perceive media without complex API integrations or middleware.
MiniMaxApr 13MiniMax Music 2.6 introduces structural control that follows specific narrative beats and intentional imperfection to make generated tracks sound more human. The update shifts AI music from random generation to a directed creative process with improved bass response and faster inference.
MiniMaxMar 25MiniMax launched the Token Plan, a flat-rate API subscription giving developers one key to access its full model suite — text, speech, music, video, and image. A single predictable bill replaces per-modality charges, starting at $100/year.
MiniMaxMar 23MiniMax released an open-source skills repository for AI coding agents under MIT license. The collection includes curated skills for iOS and Android development, document file editing, and GLSL shaders that plug into compatible coding tools for structured, production-quality guidance.
MiniMaxMar 19MiniMax released M2.7, their first model that actively participated in its own training evolution. It achieves SOTA on SWE-Pro (56.2%) and matches Claude Sonnet 4.6 in OpenClaw for agentic coding tasks, with API and MiniMax Agent access available now.
Frequently asked questions