Moonshot, a company backed by Chinese e-commerce giant Alibaba, has unveiled a new artificial intelligence model called Kimi, claiming it outperforms models from OpenAI and Anthropic in programming tasks.
According to CNBC, the Kimi K2 model is an affordable, open-source large language model developed by Moonshot. The company is adopting a strategy similar to that of DeepSeek in an effort to challenge some of its Western competitors. OpenAI recently postponed the release of its own open-source model due to security concerns.
Kimi K2 Delivers Advanced Performance in Programming
One of the Kimi K2 model’s main strengths is its ability to program applications. Both OpenAI and Anthropic have also focused their efforts on this domain. Anthropic recently introduced its Opus 4 model for AI-assisted programming.
In a statement shared on social media, Moonshot claimed that Kimi K2 has outperformed Claude Opus 4 in two benchmarks and has shown overall better performance than GPT-4.1 based on several industry metrics.
Unlike ChatGPT or Claude, which require monthly subscriptions to access their latest AI models, the new K2 model is available to users for free through the Kimi app and browser interface.
However, according to the Kimi website, the service charges $0.15 per one million input tokens and $2.50 per one million output tokens. Tokens are a unit of measurement for the amount of data processed by AI models.
Moonshot also announced on GitHub that developers are free to use K2 in any way they wish. The only condition is that if their product or service exceeds 100 million monthly active users or generates $20 million in monthly revenue, they must display the name “Kimi K2” in the user interface.
