Kimi Releases 2.8T Flagship K3.1 Pro, Tops SWE Marathon with a Score of 51.7
Kimi has released its multimodal flagship model K3.1 Pro, with 2.8T total parameters and 104B active parameters (MoE), and a standard 1M-token context window. The author says it tops the SWE Marathon leaderboard with a score of 51.7, overtaking Opus 5.5 and GPT-6 Astra, and the company has also promised to open-source the full weights by November 4. K3.1 Pro and K3.1 Pro Highspeed, which is 2–3× faster, are now live on the Kimi web app, Kimi Work, Kimi Code and the Kimi API.
Why read this
SelectedAI score 84
Kimi K3.1 Pro, with 2.8T total parameters and a 1M context, benchmarks against closed-source flagships and promises to open-source the full weights by November 4, giving readers a basis to judge the competitive position of the open-source camp.
Topics
View original postIntroducing Kimi K3.1 Pro: our most capable 3T-class open-weight model. 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal. 🔹 Full-project coding: games in Unreal, Unity & Godot. 🔹 Excels at parametric 3D modeling in Blender & CAD. 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%. 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks. Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on https://t.co/85jA71X9gw, Kimi Work, Kimi Code, and the Kimi API. Open weights by Nov 4, 2026. 🔗 API: https://t.co/XCrgjXAqMw 🔗 Tech blog: https://t.co/5iPP0MBej6
View original
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model. 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal. 🔹 Full-project coding: games in Unreal, Unity & Godot. 🔹 Excels at parametric 3D modeling in Blender & CAD. 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%. 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks. Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on https://t.co/85jA71X9gw, Kimi Work, Kimi Code, and the Kimi API. Open weights by Nov 4, 2026. 🔗 API: https://t.co/XCrgjXAqMw 🔗 Tech blog: https://t.co/5iPP0MBej6
Body · AI translation
Damn! This is explosive — Kimi just surprise-released its 2.8T flagship K3.1 Pro, scoring 51.7 on SWE Marathon to take first place, overtaking Opus 5.5 and GPT-6 Astra head-on!
And this benchmark scorecard just released has punched through more than half of the entire closed-source large model camp's defenses🤯..
Officially, the new multimodal flagship K3.1 Pro has been unveiled: total parameter count stacked straight to 2.8T, with a 1M native context window as standard.
But in my view, the 2.8T figure may just be surface muscle; what's truly lethal in the whole announcement is two things: First, in the most valuable code and long-horizon agent tests, it went head-to-head with Opus 5.5 and even overtook GPT-6 Astra.
Second, the official statement promises in black and white that full weights will be fully open-sourced before November 4.
Many people look at benchmarks and only stare at the overall leaderboard for the excitement, but carefully breaking down this benchmark table, three key battles directly determine the direction of subsequent engineering deployment: