AI Market Share Shakeup: OpenAI & Google Rise as Anthropic Falls in 2025
The AI market share landscape is undergoing significant shifts in 2025, according to new data from Poe. OpenAI's GPT-4o maintains dominance with 35.8% message share, while Google's Gemini 2.5 Pro gains traction. Meanwhile, Anthropic's Claude models have seen a 10% decline in usage share. The report highlights specialized reasoning capabilities as the new competitive battleground, growing from 2% to 10% of all text messages on Poe since January.
Key Market Shifts and Emerging Trends
Poe's comprehensive analysis of user behavior across 100+ AI models reveals rapid changes in the AI market share distribution. OpenAI's newer GPT-4.1 family captured 9.4% share within weeks of launch, while Google's Gemini 2.5 Pro achieved approximately 5% message share. These gains came largely at Anthropic's expense, though Claude 3.7 Sonnet has begun replacing its predecessor in user preference.
The most striking development is the surge in specialized reasoning model adoption. Gemini 2.5 Pro leads this category with 31% share, followed by Claude's reasoning-specialized models. OpenAI has released five reasoning models (o1-pro through o4-mini) in 2025 alone, demonstrating the intense competition in this high-value segment.
Multimedia AI Competition Intensifies
Beyond text generation, the visual AI arena shows Google's Imagen 3 growing from 10% to 30% share, challenging Black Forest Labs' FLUX models (35%). OpenAI's GPT-Image-1 achieved 17% share within two weeks of launch. In video generation, Chinese lab Kuaishou's Kling models captured 30% share, disrupting Runway's early lead.
ElevenLabs maintains audio generation dominance with 80% share, though newcomers like Cartesia and Unreal Speech are gaining traction in niche applications. These shifts reflect a broader market evolution where multimedia capabilities are becoming as strategically important as text generation.
Frequently Asked Questions
Which AI model currently has the largest market share?
OpenAI's GPT-4o maintains the largest overall share at 35.8% of text generation usage on Poe, though its dominance is being challenged by newer models.
Why are reasoning models gaining popularity?
Reasoning models demonstrate superior capability for complex tasks, with users willing to accept higher costs and processing times for improved precision in problem-solving.
How has Anthropic's market position changed?
Anthropic's Claude models saw a 10% absolute decline in usage share, though Claude 3.7 Sonnet is gaining traction over previous versions.