Alibaba’s AI beats ChatGPT, Claude On Coding Performance — That Too At A Fraction Of Cost

Kimi K2 AI News: China’s latest competitor to OpenAI’s ChatGPT is stepping up with coding capabilities with the launch of a new large language model (LLM). Moonshot, an Alibaba-backed startup, quietly unveiled its Kimi K2  late Friday night. Kimi K2 is a low-cost, open-source AI model, following a playbook similar to China-based DeepSeek, which disrupted the AI industry in January.

While most US tech giants have been reluctant to go fully open-source, Meta and Google have taken limited steps in that direction. Moonshot, however, is offering developers free access to Kimi K2’s source code.

Just hours after Moonshot’s announcement, OpenAI CEO Sam Altman revealed on Saturday that OpenAI’s first open-source model would be delayed indefinitely due to safety concerns. 

Moonshot’s Kimi K2: Coding Power at Lower Prices

One of Kimi K2’s key features is its ability to write computer code, a function that has caught the attention of businesses looking to cut costs or reduce staffing by automating coding tasks.

Anthropic, one of OpenAI’s rivals, focused on coding with its Claude Opus 4 model, which launched in late May.

Moonshot claims Kimi K2 outperforms both Claude Opus 4 and OpenAI’s coding-focused GPT-4.1 model on several industry benchmarks. The announcement was made through posts on social media platform X and GitHub.

Read More: SpaceX Invests $2 Billion In Elon Musk’s AI Company xAI To Challenge OpenAI And Google

What Makes Moonshot’s Kimi K2 Different

Cost is a major factor in Kimi K2’s appeal. Unlike ChatGPT and Claude, which charge monthly subscriptions for access to their latest models, Kimi K2 is free to use through Kimi’s app and browser interface.

For developers and businesses looking to integrate Kimi K2 into their products, Moonshot is charging just 15 cents per 1 million input tokens and $2.50 per 1 million output tokens. Tokens are the units used to measure data processing in AI models.

In comparison, Claude Opus 4 charges $15 per million input tokens and $75 per million output tokens. That makes Claude 100 times more expensive for input and 30 times more expensive for output.

OpenAI’s GPT-4.1 charges $2 for input and $8 for output per million tokens.

Moonshot has set minimal conditions for developers. If a commercial product or service using Kimi K2 has more than 100 million monthly active users or generates $20 million in monthly revenue, the product must display the label “Kimi K2” on its user interface.

Initial Reviews of Moonshot’s Kimi K2

Initial reviews of Kimi K2 on both English and Chinese social media have been largely positive. However, some users have reported instances of AI hallucinations, where the model generates false information, which is a common issue in generative AI.

Still, some in the tech community are already adopting Kimi K2 for real-world use.

Moonshot has a history of open-sourcing its previous AI models. The company’s chatbot surged in popularity early last year as a Chinese alternative to ChatGPT, which is not officially available in the country.

Since then, the Chinese market has become crowded with similar offerings from ByteDance and Tencent, while Baidu has revamped its core search engine with AI tools.

Also Read: Instagram Now Allows Search Engines Like Google To Index Public Posts: What This Means For SEO