AI Showdown: ChatGPT Losing Ground to Competitors

As of the fall of 2025, it’s clear that OpenAI and ChatGPT no longer dominate the AI market as they once did. The successive and somewhat disappointing launches of GPT-5 and its variants have let users down, while Google has made significant strides with Gemini 3 and Anthropic has continued to attract attention with Claude. Recent web traffic data confirms this trend, with Gemini experiencing rapid growth as ChatGPT significantly slows down. “Google has the models, the resources to enhance them, and now the means to deploy them on a large scale,” summarized The Verge in mid-January.

To gauge the extent of this shift, we analyzed nearly two years of data from LMArena. This system ranks AI models based on their performance and relies on blind user votes. Since April 2024, BDM has been compiling these results monthly, and the changes are striking. Although OpenAI remains a constant presence in the rankings with its various GPT versions, the company led by Sam Altman has gradually relinquished its top position to competitors, notably Google.

Editor’s Note: LMArena provides a weekly snapshot of the best current AI models based on user feedback. BDM has opted to share these monthly snapshots with our readers since the ranking system was introduced, using them to create, nearly two years later, an analysis of how major AI model publishers have evolved.

OpenAI, a Leader Increasingly Challenged

From April 2024 to December 2025, OpenAI established itself as the dominant force in AI and naturally remained the most frequent in LMArena’s monthly top 10 rankings. With 80 appearances over 21 months, the company significantly outperformed its competitors: Google (Gemini) with 56 appearances, Anthropic (Claude) 21, xAI (Grok) 15, and the Chinese firm DeepSeek 10. Only eight other publishers managed to make rare appearances in the rankings, including Alibaba (4 appearances), Meta (3), and Tencent (1).

The omnipresence of OpenAI, which never left the top 8, first reflects a technical reality. For nearly all of 2024, GPT models were simply superior to the competition. OpenAI, beyond its high-performing models, capitalized on the significant lead it had gained with the launch of ChatGPT at the end of 2022. In contrast, Google took time to enter the fray, often stumbling, while Anthropic did not have the same exposure for its Claude models.

However, the following year saw increasingly fierce competition emerge. Anthropic expanded its iterations of Claude 3 and then Claude 4, successfully establishing itself as a reliable alternative for certain applications, especially in writing tasks. xAI, despite some missteps with its AI, launched Grok, which started to climb the ranks. Similarly, DeepSeek, a Chinese player previously unknown to the general public, made a notable entry into the race. This rising competition gradually eroded OpenAI’s technical advantage, transforming what looked like a monopoly into a truly competitive market.

Google and Anthropic Overturn the Hierarchy

While OpenAI remains a frequent presence in the rankings, the company is no longer able to secure the top spots. Between November 2024 and December 2025, Google claimed the top spot nine times, compared to seven for OpenAI throughout the period. Early in 2025, xAI took two top spots, and by the end of the year, Anthropic claimed three. However, it’s the evolution over time that reveals the extent of the shift. In April 2024, OpenAI’s best model held the top ranking. By December 2025, it had plummeted to eighth place.

This gradual decline is evident month by month as the positioning of each publisher’s best models evolves. Until the fall of 2024, OpenAI proudly led the rankings. But by the end of 2024, the decline began: the company’s best model gradually slipped, overtaken by increasingly sharp competition. Google took over the top spot and solidified its presence at the forefront, while xAI and Anthropic placed their pieces and constantly disrupted the established order. OpenAI attempted a rebound in spring 2025 with the launch of GPT-4.5, but by the end of summer, it faced another significant setback. In December 2025, its best model ranked eighth, a dramatic fall for the company that had long been the absolute reference in generative AI.

The United States at the Forefront of AI, with China Lurking

The race for artificial intelligence is primarily playing out on the American continent. Between April 2024 and December 2025, models developed in the United States accounted for 88% of occurrences in LMArena’s monthly top 10 rankings. This overwhelming dominance is undoubtedly due to the concentration of tech giants, their resources, and a very permissive legislative environment: OpenAI, Google, Anthropic, and xAI all operate from the U.S. soil. However, other giants are noticeably absent. Meta, despite its colossal resources, only managed to place its Llama models sporadically, appearing just three times during the entire period. For Apple, the failure of Apple Intelligence is clear: the brand will rely on Gemini for its Siri overhaul, among other AI features.

Despite U.S. dominance, China nevertheless makes a notable showing with 11% of occurrences, an even more remarkable score given the politically tense context, characterized by technological restrictions and limited access to the most advanced chips. DeepSeek, with its open-source models, emerges as China’s spearhead with ten appearances in the top 10, demonstrating the country’s ability to develop competitive models. 01.AI, for a time, Alibaba with its Qwen models, and Zhipu AI also made a few appearances, confirming Beijing’s ambition not to leave the field open to American players.

The rest of the world struggles to make its mark in this frantic race. Canada manages to make the rankings with Cohere. Europe, on the other hand, is notably absent. Mistral, the French AI flagship often presented as the European hope, never made it into the monthly top 10s during the period. By December 2025, it had completely disappeared from the top 20, unable to compete with the American and Chinese behemoths that have financial means and computing capabilities far beyond comparison.