Open model Kimi K2.7 Code undercuts GPT-5.5 and Claude by up to 12x on price per token

Editor's take

Moonshot AI has introduced Kimi K2.7 Code, an open-weights, trillion-parameter model specifically engineered for coding tasks, significantly undercutting the per-token pricing of proprietary competitors like OpenAI's forthcoming GPT-5.5 and Anthropic's Claude 4.8.

This development is significant because it directly challenges the economic model of high-cost, closed-source LLMs in a critical application domain. By offering a performant, albeit benchmark-trailing, alternative at a drastically lower price point, Moonshot AI could democratize access to advanced coding assistance for developers and smaller organizations previously priced out of cutting-edge solutions.

Future developments to monitor include Kimi K2.7 Code's real-world adoption and its impact on the pricing strategies of major players like OpenAI and Anthropic. A key question is whether its practical utility in production environments justifies the benchmark gap, and if further open-source advancements can close that performance delta while maintaining cost leadership.