China's Moonshot AI launches Kimi K3.1 Pro; plans full model weights release by November

SynopsisChinese artificial intelligence (AI) company Moonshot AI has unveiled Kimi K3.1 Pro, a 3-trillion-parameter open-weight AI model featuring a 1-million-token context window, alongside a high-speed variant, Kimi K3.1 Pro Highspeed, according to a company statement.Listen to this article in summarized formatReutersFILE PHOTO: Moonshot AI logo is seen in this illustration taken, September 5, 2026.

REUTERS/Dado Ruvic/Illustration/File PhotoChinese artificial intelligence (AI) company Moonshot AI has unveiled Kimi K3.1 Pro, a 3-trillion-parameter open-weight AI model featuring a 1-million-token context window, alongside a high-speed variant, Kimi K3.1 Pro Highspeed, according to a company statement.According to the release, both Kimi K3.1 Pro and its high-speed variant are now available across Kimi.ai, Kimi Work, Kimi Code and the Kimi API, while the company plans to release the full model weights by November 4."We're introducing Kimi K3.1 Pro, the first model in our new K3.1 series and our most capable 3T-class open-weight model to date, with a 1-million-token context window.

It delivers substantial improvements across coding, 3D creation, financial research, and other agentic capabilities," it said.According to the company, Kimi K3.1 Pro is an upgraded version of Kimi K3, developed using significantly greater computing resources during post-training. Also, "K3.1 Pro covers the full range of coding work, from quick fixes in real codebases to full-project builds," and is capable of end-to-end delivery of complete financial workflows.Furthermore, Moonshot AI has optimised its reinforcement learning (RL) framework, long-context processing, training efficiency and infrastructure to improve the model's performance across hundreds of application areas, including coding, autonomous AI tasks and knowledge-based operations, it said.As per the company, K3.1 Pro delivers performance comparable to leading proprietary AI models across several benchmarks.On Terminal-Bench 4.0, which evaluates coding agents on real-world terminal-based tasks, the model scored 65.7 per cent, surpassing GPT-6 Astra's 58.2 per cent and coming close to Claude Opus 5.5's 66.4 per cent.Meanwhile, it noted, "On Kimi Code Bench 3.0 Long-horizon, our internal benchmark that tests coding agents' multi-step execution, tool use, and end-to-end problem solving across diverse engineering domains, K3.1 Pro scores 61.2%, above K3's 55.95."Apart from this, Kimi K3.1 Pro Highspeed "delivers 2-3x the speed of K3.1 Pro for more responsive interactions," it said.Kimi K3.1 Pro is now available across the company's platforms, including Kimi.ai, Kimi mobile apps, Kimi Work and Kimi Code, as well as through the Kimi API, the company added. ...moreElevate your knowledge and leadership skills at a cost cheaper than your daily tea.KYB and KYC: Business behind the wealth creator called ‘Divi’s Lab’ – Part 2, concludingKYB and KYC: Business behind the wealth creator called “Divi’s Lab” – Part 1The global business of promising moreOne company, 2 words.

Bond yield and CDS: Will they decide the fate of global markets?Business of Odd Things: Inside an industry that farms fish that nobody is meant to eatWar's second-order effects are becoming visible in critical sectorsUS mid-term elections – Part 2: What does it mean for Trump to lose control of Congress?Homebuyers wait as NCLT panel picks holes in unfinished Jaypee projectsWhat is the real impact of 13x jump in cash transfers to women?Airlines Q2 turbulence: Fewer passengers, rising fares and a crude shockSubscribe Now