The news: OpenAI and Google have gained share in key AI categories, per Poe’s Spring 2025 AI Model Usage Trends, which tracked usage across 100+ models by message volume.

GPT-4o and Gemini 2.5 Pro have taken the lead in both text generation and reasoning usage. Meanwhile, Anthropic’s Claude and DeepSeek-R1 lost momentum following a surge of usage at the start of the year. 

AI’s shifting sands: The highly competitive AI market continues to shift dramatically as new models ramp up performance. The share of user messages sent to reasoning models—which can solve complex questions but take more time—grew from 2% to 10% of overall usage in just five months.

More user trends:

• OpenAI’s GPT-4o captured a 35.8% message share while GPT-4.1 added 9.4%.
• Google’s Gemini 2.5 Pro leads the reasoning category with 31% message share.
• Anthropic’s Claude seems to be stalling with a 10% usage drop between January and May as users shifted to newer models.
• DeepSeek-R1’s message share declined from a peak of 7% in mid-February to 3% by the end of April.

Image and video AI leaders: In terms of AI for image and video content creation, Google’s Imagen 3 is challenging incumbents by growing its 10% share to 30% share from January to May. GPT-Image-1 and FLUX remain strong, but their dominance is shrinking. 

Other standouts:

• For AI video creation, Kuaishou’s Kling upended Runway’s early lead, hitting 30% share just weeks after launch.
• OpenAI’s GPT-Image-1 captured 17% image share in just two weeks after its late April launch.
• Runway’s market share fell dramatically from 60% to 20% between January and May.

Key takeaway: Google’s AI offerings are quickly gaining traction across various categories, indicating that while it is coming from behind, it’s only a matter of time before it emerges as the industry leader, provided it can sustain innovation.

Google’s large install user base and wide range of tools and services give it a competitive advantage in adoption as well as in user training data.