🚀 Moonshot открыла веса Kimi K3.1 Pro Китайская Moonshot AI опубликовала полные веса своей новой модели Kimi K3.1 Pro на Hugging Face. Ранее она была доступна только через API и приложение. Что интересного: ➖ 2,8 трлн параметров, из которых на каждый токен активны 104 млрд. ➖ Контекстное окно — 1 млн токенов. Модель работает с текстом и изображениями. ➖ Веса опубликованы в сжатом формате MXFP4. Весь репозиторий занимает около 1,56 ТБ. ➖ Модель ориентирована прежде всего на программирование и работу ИИ-агентов. По независимым тестам Vals AI, Kimi K3.1 Pro набрала 62,52 балла в Vals Index, опередив DeepSeek V4.1 Flash с результатом 61,26. В Terminal-Bench модель показала 63,64%, хотя до Claude Opus 5.5 с её 66,4% пока немного не дотягивает. Источник
From the original source:
Telegram — GPT Time | Нейросети | ИИ
Read full article at Telegram — GPT Time | Нейросети | ИИMoonshot revealed the weights of Kimi K3.1 Pro
🚀 Moonshot has revealed the weights of the Kimi K3.1 Pro Chinese Moonshot AI has published the full weights of its new model Kimi K3.1 Pro on Hugging Face. Previously, it was only available through the API and application. What's interesting: ➖ 2.8 trillion parameters, of which 104 billion are active for each token. ➖…
AI Summary & Key Takeaways
- 🚀 Moonshot has revealed the weights of the Kimi K3.1 Pro Chinese Moonshot AI has published the full weights of its new model Kimi K3.1 Pro on Hugging Face
- Previously, it was only available through the API and application
- What's interesting: ➖ 2.8 trillion parameters, of which 104 billion are active for each token
Moonshot revealed the weights of Kimi K3.1 Pro
Related stories
Rylis: AI assistant for iPhone that can respond without the Internet
What to do if you need an assistant, but there is no connection: in the mountains, on a highway far from the city, or in a village with poor Internet? Alexander Lavrikov, developer of the Rylis application, is working on this. It runs language models directly on the phone, and once the model is downloaded, local mode can...
GPT-6.2 Leaks Are Getting Seriously Interesting
OpenAI's next model is getting very close, and the leaks around it are getting harder and harder to ignore. Two mysterious model names appeared in OpenAI's official list of models for a few minutes and then vanished. A giant new model called Bell is reportedly sitting inside…
Anthropic has eliminated token overspending in Claude Code
✔️ Anthropic has eliminated token overspending in Claude Code Previously, Claude Code background subagents performing code reviews or tests updated their status every 30 seconds and each time the base model re-read the entire correspondence history. Users online noted that...
Thinking Machines released an AI model that independently trained another model
Thinking Machines released an AI model that independently trained another model. The former OpenAI team released their open AI model Inkling 2 Small, which was given the task of teaching Qwen3.5-4B to forecast time series. Then Inkling 2 Small acted alone: she wrote...