MiniMax debuts AI model built for long and complex coding tasks

The Chinese artificial intelligence startup MiniMax has launched its latest flagship model, M3, a system built specifically to handle long and complex coding tasks. The debut marks the company’s first major product release since it formally began preparations for an initial public offering on Shanghai’s Star Market, complementing its existing listing in Hong Kong. But the real story is not the model’s technical specs—it is the price tag. MiniMax claims M3 cuts inference costs by 95% compared to its predecessor. That is not a marginal improvement; it is a structural shift. If the claim holds in production, it threatens to commoditize AI coding agents almost overnight. Competitors in China’s already overheated AI market will have to respond, likely triggering another round of price wars that erode margins across the sector. The model can process up to 1 million tokens of data at once—five times more than its predecessor, the M2.7. This capacity allows M3 to handle entire programming projects, not just isolated snippets. Instead of merely answering questions, the model is designed to act as an AI project manager, handling multi-step software tasks and running entire workflows with minimal human supervision. MiniMax has emerged as one of China’s most commercially successful AI companies, but its strategy has never been about chasing every benchmark. Speaking at an earnings briefing in March, CEO Yan Junjie said the firm was not aiming for blanket dominance across every AI metric. That comment now reads as a deliberate positioning: MiniMax is betting on specialized, high-value use cases rather than general-purpose supremacy. The company recently introduced Mavis, a multi-agent system that allows several AI agents to operate simultaneously on a single device, each delegated to distinct tasks. M3 serves as the bedrock for this broader agentic AI strategy. The combination of ultra-low inference costs and multi-agent orchestration could make MiniMax a formidable player in enterprise automation, particularly for software development teams that need to scale without adding headcount. What a casual observer might miss is the timing. MiniMax’s IPO preparations and the M3 launch are not coincidental. The company is signaling to investors that it can compete on cost while still delivering cutting-edge capability. In a market where many AI startups burn cash on model training and offer services at a loss, MiniMax is trying to prove it can build a sustainable business around agentic workflows. The risk is that commoditization cuts both ways. If M3’s cost advantage is real, it will pressure rivals like Baidu, Alibaba, and ByteDance to slash prices on their own coding agents. That could compress margins for everyone, including MiniMax. The company’s bet is that volume and stickiness—via multi-agent systems like Mavis—will offset thinner per-unit margins. For now, MiniMax has thrown down a gauntlet. The question is whether its competitors can match the price without sacrificing quality, or whether they will retreat to higher-margin niches. The next few months will show whether M3 is a disruptive force or just another entry in a crowded field.

Chinese artificial intelligence start-up MiniMax has unveiled its latest flagship AI model, M3, designed to anchor the company’s push into coding agents and automated workflows.

MiniMax’s M3 cuts inference costs by 95%, threatening to commoditize AI coding agents and intensify price wars among Chinese AI vendors.

The development adds to a wider China ai & machine learning story in which companies are being judged on execution, capital access, regulatory fit and the credibility of their regional expansion plans.

For business readers, the important question is whether this becomes an isolated announcement or part of a more durable operating pattern across customers, financing channels, partners and public-market expectations.