Chinese AI Kimi K2 proves more powerful than ChatGPT in coding and costs next to nothing

News

Translated with AI from the Russian original.

Alibaba-backed Chinese company Moonshot has unveiled Kimi K2, a new open-source language model. It not only outperforms rivals like GPT-4.1 and Claude Opus 4 in coding benchmarks, but is also available for free.

The model focuses on code generation and is already available in browsers and a mobile app. For commercial use, pricing is among the lowest on the market: $0.15 per million input tokens and $2.50 per million output tokens. This is tens, and sometimes a hundred times cheaper than counterparts from Western companies.

Kimi K2 achieved strong results in benchmark tests, outperforming Claude and GPT-4.1. The developers assert that it is ideally suited for tasks related to writing and analyzing code.

However, the model still has limitations. Tools for integration with external systems are still in development, which may hinder quick adoption in enterprise environments. Users also report rare cases of “hallucinations”—a typical problem of generative AI.

The release of Kimi K2 coincided with delays in the release of an open-source model from OpenAI. This provided the Chinese development with a chance to outpace American competitors and secure a strong position in the global AI market.

***

Cover photo author: Bloomberg. Cover photo source: CNBC

Never miss our newsAdd this site to your preferred sources to see us more oftenAdd on Google

Now on the front page