OpenAI makes GPT-4o free for all, putting real-time voice and vision in everyone's pocket

OpenAI just reset the playing field. The company’s spring update wasn’t about a GPT-5 reveal or a search engine launch, but something arguably more disruptive: the new GPT-4o flagship model is available to ChatGPT free users immediately. That’s the same ‘omni’ model that natively handles audio, vision, and text in real time, and it’s no longer locked behind a $20 paywall.

This is a genuine multimodality play. The ‘o’ stands for omni, and the live demos showed a conversational AI that can see through your camera, respond to your tone of voice, and interrupt itself mid-sentence — all with latency that feels closer to a human chat than a buffering robot. It’s a stark departure from the clunky, sequential voice mode many users are accustomed to, which relied on separate models for transcription and response.

It’s also an unsubtle shot across the bow at competitors like Google, which has been scrambling to integrate Gemini into its ecosystem. By giving away the most advanced model for free — while simultaneously refreshing the ChatGPT UI with a cleaner, more approachable design — OpenAI is aggressively chasing mass adoption. Paying users still get a capacity advantage, with five times the usage limits, but the core intelligence is now a commodity. The real question is whether this level of free access is a temporary land grab or a permanent shift in OpenAI’s business model as it scales inference costs.