OpenAI chief executive Sam Altman signalled on Saturday, 28 June 2026 that the company is actively working toward significantly expanded token allowances for users, responding to a query on X with a characteristically understated but consequential remark.

Replying to a user on X, Altman wrote: 'not quite all-you-can-eat tokens, but we are working on it' — a phrase that immediately drew attention from developers and AI enthusiasts worldwide, including a large and growing community of ChatGPT power users in India.

Context

Token limits have been a persistent friction point since OpenAI launched ChatGPT in 2022. The platform introduced rate limits tied to token consumption across both free and paid subscription tiers, a mechanism designed to manage surging inference demand against constrained compute capacity. For millions of users — particularly API developers and enterprise clients — these caps have represented a ceiling on productivity.

The phrase 'all-you-can-eat tokens' refers to a hypothetical flat-rate or unlimited-access model, where users would not face hard cutoffs on how much they can interact with a model in a given period. Altman's reply stops short of confirming such a plan, but acknowledges it as a direction the company is pursuing.

Policy Backdrop

Token-based metering has become the industry standard for controlling access to frontier AI models. OpenAI and peer laboratories have repeatedly revised usage caps and pricing structures to balance user demand with the real-world costs of running large language models at scale. Each revision has carried significant downstream consequences for developers who build products on top of these APIs.

In India, where a fast-growing developer ecosystem relies heavily on OpenAI's API, any shift in token pricing or access policy is watched closely. Startups and independent developers operating on tight budgets are particularly sensitive to changes in how consumption is metered and billed.

Stakeholders and Impact

The most directly affected parties are ChatGPT subscribers on paid tiers and the large community of API developers who integrate OpenAI's models into applications. For consumer users, expanded token limits would mean longer, more complex conversations without hitting usage walls mid-session. For API developers, it could translate into lower per-unit costs or more predictable flat-rate billing.

Indian technology startups, many of which have built generative AI products on OpenAI's infrastructure, stand to benefit significantly if the company moves toward more generous or restructured access tiers. The broader pattern across the AI industry — with multiple labs experimenting with higher-volume and flat-rate plans — suggests competitive pressure is accelerating these changes.

What's Next

Altman's post does not specify a timeline or the precise form that expanded token access might take. Observers will be watching for formal announcements on revised API tiers, new subscription offerings, or changes to the ChatGPT Plus and Pro plans. Any such announcement would carry immediate implications for how businesses and individual users in India and globally plan their AI spending. The remark, brief as it is, signals that the question of access limits remains very much an open and evolving one at the highest levels of OpenAI.