Moonshot AI Teases Upcoming Kimi K3.1 Model
Moonshot AI released a 24-second teaser video featuring the upcoming Kimi K3.1 model running inside Kimi Code. This follows developer leaks and code commits in open-source frameworks revealing expanded reasoning intensity tiers. The update highlights Moonshot AI's continued focus on developer workflows, programming assistance, and refined reasoning control. It signals that the ecosystem is preparing for a new flagship iteration following widespread enterprise adoption of the K3 model. K3.1 expands reasoning intensity tiers from three (low, high, max) to four by adding a new medium tier, while retaining a 1 million token context window. Code repository commits also point to compatibility testing for chat encoders, tool calling, and image inputs.
## BACKGROUND
Moonshot AI is a Chinese artificial intelligence company known for developing the Kimi chatbot and large language models capable of processing extremely long contexts. SGLang is an open-source framework designed for high-throughput and low-latency inference serving of large language models and multimodal models.