Chinese AI company Moonshot AI has introduced Kimi K3.1 Pro, a new model designed to handle coding, complex tasks and large volumes of information.
The model has 2.8 trillion total parameters and a one-million-token context window. This allows it to process extensive documents, large codebases and complex projects within a single context. Kimi K3.1 Pro is currently available through Moonshot’s platforms and API.
Close to Claude Opus 5.5 on a coding benchmark
According to figures published by Moonshot AI, Kimi K3.1 Pro scored 65.7% on Terminal-Bench 4.0, compared with 66.4% for Anthropic’s Claude Opus 5.5.
Terminal-Bench 4.0 evaluates AI agents on practical tasks performed through a computer terminal. These tasks go beyond generating code and include solving problems and working within software development environments.
However, the reported scores come from Moonshot AI. Until independently verified, they should not be interpreted as proof that Kimi K3.1 Pro matches Claude Opus 5.5 across all coding tasks or real-world applications.
Why open weights matter
One of the model’s key features is Moonshot AI’s plan to release its full model weights on November 4, 2026.
If the weights become publicly available under suitable licensing terms, developers may be able to adapt the model to specific use cases and run it on their own infrastructure, provided they have the necessary computing resources. This could give businesses and developers greater control over deployment and data handling while reducing reliance on externally hosted AI services.
However, access to model weights does not automatically make deployment inexpensive or straightforward. Running a model of this scale can require substantial computing resources, and the available license will determine how it can be used.
The release of Kimi K3.1 Pro adds to competition between AI developers building models for advanced coding and other complex tasks. Its actual capabilities will become clearer once the full weights are released and independent evaluations become available.
