AI Market Shifts Reveal OpenAI and Google Lead with Specialized Reasoning Models Rising
Poe's latest report reveals significant shifts in AI usage from January to May 2025, with OpenAI and Google strengthening their lead across key AI categories. Specialized reasoning models have surged, becoming a critical competitive edge. Meanwhile, Anthropic loses market share, and new entrants disrupt video and voice AI sectors. This evolving landscape underscores the importance of flexible AI strategies for businesses.
The AI landscape is rapidly evolving, as highlighted by Poe’s latest usage report covering January to May 2025. This report, based on data from Poe’s platform that provides access to over 100 AI models, offers unique insights into real user preferences and market dynamics beyond traditional benchmarks. OpenAI and Google have solidified their leadership positions in multiple AI categories, while Anthropic’s market share has declined.
In text generation, OpenAI’s GPT-4o remains dominant with a 35.8% usage share, and its newer GPT-4.1 models quickly captured 9.4% shortly after launch. Google’s Gemini 2.5 Pro also gained traction, achieving around 5% share. Anthropic’s Claude models experienced a 10% drop, with Claude 3.7 Sonnet replacing earlier versions but still maintaining a notable user base.
A standout trend is the rapid rise of specialized reasoning models, which have grown from 2% to 10% of all text messages on Poe since early 2025. These models excel in handling complex tasks with greater precision, a capability increasingly valued by early adopters despite higher costs and processing times. Google’s Gemini 2.5 Pro leads this segment with 31% usage, surpassing Claude’s reasoning models. OpenAI has aggressively expanded its reasoning offerings with multiple new models released in early 2025.
Hybrid reasoning models, which dynamically adjust reasoning levels during conversations, have emerged but currently represent a small fraction of usage. This shift toward reasoning capabilities signals a maturing AI market where raw text generation is becoming commoditized, pushing providers to differentiate through advanced, high-value features that justify premium pricing.
In image generation, Google’s Imagen 3 has grown from 10% to 30% market share, challenging Black Forest Labs’ FLUX models, which hold about 35%. OpenAI’s GPT-Image-1 quickly captured 17% usage shortly after release. This three-way competition marks a shift from earlier dominance by Midjourney and Stable Diffusion, driven by rapid improvements in image quality, prompt fidelity, and rendering speed. Enterprise adoption is accelerating, especially in marketing and creative workflows.
The video generation market has seen disruption from Chinese newcomer Kuaishou’s Kling models, which captured 30% usage with Kling-2.0-Master alone reaching 21% within weeks. Google’s Veo 2 holds 20%, while early leader Runway’s share dropped from 60% to 20%. Video generation remains computationally intensive but is poised for broader adoption as costs fall and quality improves, with early enterprise users in advertising, social media, and education.
In audio generation, ElevenLabs dominates with 80% of text-to-speech requests, thanks to its early market entry and extensive voice library. However, new entrants like Cartesia, Unreal Speech, PlayAI, and Orpheus are gaining ground by targeting niche markets such as podcasting, multilingual voices, and emotional voice effects. The audio AI market is expected to grow significantly as text-to-speech approaches human-like quality and finds applications in customer service, accessibility, and content creation.
For businesses, the rapidly shifting AI model landscape presents both opportunities and challenges. Maintaining flexibility through provider-agnostic evaluation pipelines is critical, as the best-performing models can change quickly. Platforms like Poe that offer access to multiple models enable users to compare outputs and adapt strategies efficiently. The growing emphasis on reasoning capabilities suggests a shift toward precision and reliability in AI deployments.
Enterprises are increasingly adopting a portfolio approach, leveraging different AI models for varied tasks while retaining the ability to switch providers as capabilities evolve. Multimedia AI capabilities—spanning text, image, video, and audio—are becoming equally important, signaling a more balanced AI ecosystem. Success will favor organizations that focus on use-case-specific evaluations and build infrastructure for rapid adoption of meaningful innovations.
In summary, Poe’s 2025 AI usage report highlights a dynamic market where leadership is continuously contested. OpenAI and Google lead with innovative reasoning and multimodal models, while new entrants disrupt specialized niches. Businesses must stay agile and informed to harness AI’s evolving capabilities effectively and maintain competitive advantage.
Keep Reading
View AllCRISPR Patent Battle and AI Police Tech Challenge Legal Boundaries
US court reopens CRISPR patent dispute; police use AI to bypass facial recognition bans; key tech and policy updates.
LegoGPT AI Transforms Text into Physically Stable Lego Designs
Discover LegoGPT, an AI that converts text into stable Lego builds with 98% accuracy, revolutionizing creative construction.
Sony Xperia 1 VII Brings AI Video Tools and Audio Upscaling to Premium Android Market
Sony Xperia 1 VII offers AI-powered video shooting, audio upscaling, and premium features at a high price point.
AI Tools Built for Agencies That Move Fast.
QuarkyByte offers in-depth analyses and real-time insights into AI market trends like those revealed by Poe. Discover how to leverage emerging reasoning models and multi-modal AI capabilities to optimize your AI strategy and stay ahead in this fast-evolving ecosystem. Explore QuarkyByte’s tailored solutions to navigate AI’s competitive landscape effectively.