a16zWhere does consumer AI stand at the end of 2025?
At a glance
WHAT IT’S REALLY ABOUT
Consumer AI in 2025: ChatGPT leads, multimodal virality reshapes competition
- ChatGPT remained the clear consumer leader in 2025 (roughly 800–900M WAU) with strong brand “Kleenex” effects, while Gemini accelerated recently driven by viral multimodal model launches like NanoBanana.
- The biggest consumer growth catalysts shifted from text chat to image/video breakthroughs—especially realism, controllability, and multimodal reasoning—plus surprising enablers like search integration and audio-in-video.
- Product execution and onboarding nuances (templates, trending prompts, frictionless first success) increasingly determine whether users try and stick with new capabilities, even when model quality is high.
- Several “new primitives” (connectors, proactive assistants like Pulse, browser agents like Comet) look strategically important but still feel unreliable or mis-executed, creating room for startups to win with opinionated UX.
- For 2026, the group expects more “anything-in/anything-out” multimodality, enterprise-to-consumer spillover via workplace adoption, rising app-generation efforts, and continued opportunity for startups—especially outside pure text-in/text-out assistants.
IDEAS WORTH REMEMBERING
5 ideasChatGPT’s lead is reinforced by brand and habit, not just model quality.
The panel cites low multi-homing (only ~9% pay for more than one assistant) and that <10% of ChatGPT users even visited other major providers for much of the year, suggesting inertia and “default assistant” behavior matter as much as features.
Gemini’s clearest wedge is best-in-class multimodal virality, not generic chat.
Viral launches like NanoBanana (and Veo for video) pull users into Google surfaces even if they don’t think of Gemini as their primary assistant, and Android distribution meaningfully boosts mobile scale versus iOS.
Modern image/video differentiation is shifting from aesthetics to realism + reasoning + accuracy.
Beyond “style,” users now expect correct physical details, multi-input coherence, and even factual accuracy (e.g., product or historical fidelity), which sometimes requires search integration rather than just better generation.
UX templates and guided starting points are becoming the growth lever.
They contrast Gemini’s blank-slate prompt box with ChatGPT’s trending/template-led flow that nudges users into a first successful generation, then keeps them creating via character consistency and iterative suggestions.
Some under-hyped features (connectors, proactive nudges) are strategically huge but need reliability.
Connectors to email/calendar/docs and proactive products like Pulse hint at an “everything/workspace” direction, yet inconsistent performance and unclear everyday value currently limit retention.
WORDS WORTH SAVING
5 quotesFor most of the year, less than 10% of ChatGPT users even visited another one of the big LLM providers like Gemini
— Olivia Moore
ChatGPT is like the Kleenex of AI.
— Bryan Kim
These are product nuances that I, I think makes people actually take the first step- to generate it.
— Bryan Kim
I think it's because it's not the core competency of these companies anymore- to build opinionated standalone consumer UI.
— Olivia Moore
I genuinely believe that the models have gotten to the level of quality that you can build a real scalable app on top of them, and so the hope is 2026 will be a huge year for consumer builders.
— Justine Moore
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome