MoonshotAI: Kimi K2.7 Code

Kimi K2.7 Code is a text-and-image input model from MoonshotAI with a 262,144-token context window that matches its maximum output length, making it suited for long-document workflows. It supports tool use and reasoning, and its name signals a coding orientation, which the benchmark data backs up: it scores a perfect 100.0 on the AI Index coding subset while sitting at 48.8 on agentic tasks. For comparison purposes, the model's blended score of 74.3 spans only three benchmarks, so treat that figure as directional rather than comprehensive. At $0.612 per million input tokens and $3.069 per million output tokens, it sits in a mid-range price tier. Teams prioritizing code generation over general or agentic workloads are the clearest fit, while those needing structured output support should note that capability is unconfirmed from available data.