Artificial intelligence is moving beyond simple chatbots. The latest models are being built to reason through difficult problems, write and debug code, analyse information and carry out longer, more complicated tasks with less human guidance.

That rapid progress has also turned AI into a global competition.

An Artificial Analysis ranking shows that the United States and China are now the two main centres of advanced AI development. Every company in the top 10 shown on the graphic above comes from one of two countries: five American companies and five Chinese companies.

At the top is US-based Anthropic, whose Claude Fable 5 model recorded the highest score in the ranking. OpenAI's GPT-5.6 Sol ranks second, while China's Moonshot AI, with its Kimi K3 model, ranks third.

SpaceXAI's Grok 4.5 comes fourth, followed by China's Z AI with GLM-5.2. Meta, Alibaba, Google, MiniMax and DeepSeek make up the rest of the top 10.

Artificial Analysis uses its Intelligence Index to compare models across a range of tests covering areas such as reasoning, coding, science , knowledge and complex problem-solving. The results are combined to give an overall score.

That means one model may perform particularly well at coding while another is stronger at reasoning or other tasks. The ranking is therefore best understood as a comparison of overall performance rather than a definitive answer to which AI model is the best.

The leaderboard can also change quickly. AI companies are releasing new and updated models at a rapid pace, and a new release can change the rankings within days or weeks.