Chinese AI developer MiniMax debuts M3, a new coding model that it says rivals Claude Opus 4.7, costing $0.12 per 1M input tokens, compared with $5 for Opus 4.7
MiniMax has been iterating quickly through M2.1, M2.5, and M2.7, combining coding-focused upgrades with claims of sharply lower inference pricing. It also previously open-sourced MiniMax-M1 and has signaled greater investment in open-source tools and community support.
M3 extends that arc by pairing a claimed high-end coding-performance comparison with a far lower input-token price. The move arrives as MiniMax is preparing for a Hong Kong listing and is reportedly developing a larger M3 Pro model that it may release as open source.
First-order effects
Developers and enterprises evaluating coding models gain a substantially cheaper MiniMax option, conditional on M3 delivering the claimed capability in their workloads.
MiniMax strengthens its commercial pitch around low-cost model access while creating a clearer performance-and-price comparison with Opus 4.7.
Second-order effects
Premium coding-model providers face more pressure to demonstrate differentiated quality, reliability, tooling, or economics rather than compete solely on benchmark-adjacent capability claims.
Lower-priced coding inference can make higher-volume AI coding workflows more viable for customers, while increasing the importance of testing models on production tasks rather than relying on vendor comparisons.
Third-order effects
If MiniMax sustains capability gains alongside falling prices and follows through on an open-source M3 Pro, the coding-model market could segment further between premium proprietary services and low-cost or open alternatives.
The pattern would shift competitive advantage toward distribution, developer tooling, and the ability to finance rapid model iteration—not merely access to a capable base model.
The trend: M3 is part of the broader race to commoditize capable coding intelligence through faster release cycles, lower inference prices, and a mix of proprietary and open-source distribution.
🚀 We're launching MiniMax M3 from @MiniMax_AI on Novita AI as a Day-0 API launch partner. The first open-weights model to combine: • Frontier coding and agent capabilities 59.0% SWE-Bench Pro, 66.0% Terminal Bench 2.1, 74.2% MCP Atlas • Up to 1M tokens of context Enabled by [vide…
Several sleepless nights later, M3 is finally here. Coding frontier. 1M context. Native multimodal input. The first open-weights model to bring all three together. Hope you all like it 😎 p.s. M3 has already submitted a lot of PRs into MiniMax Code.
I do appreciate it when Chinese AI players compare themselves to the big 3 rather than each other, since it gives better indication of where they are performance wise. Minimax does focus on coding, work & agentic flow. That's where money is. They don't pretend to do LLMs that
New open model: MiniMax M3 by @MiniMax_AI is live in the Arena! Find it across Text, Vision, Document and Code Arena: Frontend. Bring your toughest prompts and vote. Scores incoming soon! [image]
MiniMax-M3 by @MiniMax_AI is now live on Venice. The first open-weight model to deliver frontier coding and agentic performance with native multimodality. Built for long-horizon, complex workflows. Available anonymously.
🔥API Pricing & Promotion - 50% off standard usage (≤512K context) during the first 7 days - Priority access available through: api@minimax.io - Self-serve access for all users coming in the next few days [image]
There's going to be a point where intelligence saturates meaningfully to be useful for humans, in which case open source will catch up to the closed frontier. Then, the battles will be fought elsewhere. Serving cost mostly probably?
.@MiniMax_AI M3 model is available on Ollama's Cloud! In partnership with MiniMax, the M3 model on Ollama's Cloud is US-based with zero data retention. Try M3 on coding and agentic tasks: Claude Code: ollama launch claude —model minimax-m3:cloud Codex: ollama launch
MiniMax-M3 is live on OpenRouter! A frontier-class open-weight model that combines a 1M-token context window, frontier coding and agentic performance, and native multimodality (image & video) in one model. [image]
Minimax Team has texted us at - 8 PM on a Sunday - 11 AM on a Wednesday I genuinely do not know which timezone these people they work in and that's how I know they will succeed. Kudos for the Minimax 3 release
Minimax M3 is now live in Hermes Agent on Nous Portal, OpenRouter, and Minimax Direct Providers! No need to hermes update, it should appear in your model picker automatically! [image]
MiniMax M3 scores 90.3% GPT 5.5 Scores 92.4% Just a 2.1% gap now at @convex. Incredible to see the open-source models closing in. The future of AI is open. 🤖💪 [image]