Moonshot AI's Kimi K3.1 Model Reportedly Surfaces Online
Developers have reportedly found references to Moonshot AI's Kimi K3.1 model in its API registry, with indications of a context window up to 1 million tokens. The company has not confirmed these details.
References to a new iteration of Moonshot AI's Kimi model, designated K3.1, have reportedly been discovered by developers within the company's API model registry. The surfaced information suggests the potential for a context window of up to 1 million tokens, a significant increase over current capabilities.
In addition to the large context window, the apparent details also hinted at various reasoning levels for the model. However, Moonshot AI has not yet provided any official confirmation of these specifications or released documentation for Kimi K3.1.
The "kimi-k3-1" identifier and accompanying preview, spotted by developers, indicate ongoing development for the AI model. A context window of this magnitude would enable the model to process and understand significantly longer inputs, such as extensive documents or lengthy conversations.
While the implications of such an advanced model are substantial, the information remains unconfirmed by Moonshot AI. The company has not announced a release date or further details regarding Kimi K3.1, leaving the industry to await official statements.