OpenAI launches GPT-5.3-Codex as ChatGPT ads begin testing

The ads, which are now appearing for a small subset of users, are designed to be clearly labeled and will not influence the AI’s responses, according to OpenAI’s announcement. The initial test is limited to logged-in adult users on the Free and Go tiers in the United States. The company emphasized that the advertising program is built with strong privacy protections and will not be eligible to appear near conversations on sensitive or regulated topics like health, mental health, or politics, as reported by TechMeme.

This move represents a significant, and perhaps ironic, shift in strategy for OpenAI CEO Sam Altman. As noted in a Business Insider AI report, Altman has historically expressed a dislike for advertising-based business models. The article by Peter Kafka points out that Altman is now leveraging the massive user base of free ChatGPT users, transforming them into a new potential revenue stream. This pivot suggests a pragmatic approach to funding the immense computational costs of providing free access to one of the world's most popular AI platforms.

The introduction of ads is a clear step towards diversifying OpenAI's revenue beyond its subscription-based ChatGPT Plus plan. It directly monetizes the vast audience that uses the free version, a group that until now has been a cost center. The careful framing of the announcement—highlighting answer independence, user control, and privacy—appears to be a deliberate effort to mitigate potential user backlash against a traditionally intrusive monetization method. OpenAI is attempting to prove that ads and a quality user experience are not mutually exclusive.

The timing of this test is particularly notable as it coincides with the rollout of a new, more powerful AI model for developers. While the consumer-facing ChatGPT begins seeing ads, the underlying technology for coders is getting a significant speed boost. According to Hacker News Newest, OpenAI has launched GPT-5.3-Codex, which delivers up to 25% faster performance on agentic coding tasks. This new benchmark is now being integrated into developer tools like GitHub Copilot, highlighting the company's dual-track approach of serving both enterprise and consumer markets simultaneously.

For users, the immediate change will be subtle. The core promise of ChatGPT remains intact: the answers won’t change because of an advertiser’s influence. The experience is not yet a free-for-all; the initial test is a controlled experiment to gauge user reaction and ad effectiveness within a unique conversational environment. How users respond to this new element in their chat interface will likely determine how quickly, and how broadly, OpenAI decides to expand the program. The company is walking a tightrope, balancing the need for sustainable revenue against the preservation of its product’s clean, utility-first feel.