OpenAI's Pro page lists three GPT-5.6 models as "Unlimited*"

Why it matters

OpenAI is selling access across ChatGPT, Work, Codex and its API, each with distinct usage rules. A single "Unlimited*" label can obscure the difference between model access and metered agent work.

OpenAI's Pro plan page appeared to list GPT-5.6 Sol, Terra and Luna as "Unlimited*" on September 29th, according to developer Ian Nuttall (@iannuttall). The label leaves a practical question for subscribers: unlimited in which OpenAI product, and against which usage allowance?

Nuttall, a software engineer and entrepreneur who says he builds internet businesses and AI tools, framed the listing as a possible response to a pricing change. He called Sol and Luna X-High two of his favorite models, but his thread offered an interpretation of the plan page, not a new OpenAI policy announcement. He also linked a post from Thibault "Tibo" Sottiaux, which said OpenAI planned to add subscription features the next day that would not draw on usage. The post did not name those features. Nuttall wondered whether the additions would apply beyond ordinary chat.

OpenAI's own pricing page and help documentation describe access in product-specific terms. The help page says GPT-5.6 Sol, Terra and Luna are available in ChatGPT Work for paid plans, and in Codex for Plus, Pro, Business and Enterprise users. Developers can use all three through the API. Terra and Luna are not selectable in standard ChatGPT conversations, although Luna powers the default text-chat experience for Free and Go users.

Those distinctions complicate a single "Unlimited*" label. In the pricing-page version available during reporting, OpenAI lists unlimited everyday text chats across plans and marks GPT-5 Thinking Mini as unlimited for Pro. It does not show the same unlimited marker beside GPT-5.6 Sol, Terra or Luna. The page's asterisk says usage must be reasonable and comply with OpenAI's policies. That snapshot does not establish what Nuttall saw on September 29th, but it shows why the product surface and date matter when reading a plan comparison.

OpenAI's help page separately says Work and Codex have usage and credit rules distinct from Chat. It also lists specific allowances for the higher-compute GPT-6 Pro and GPT-5.6 Sol Pro options in Chat. Those caps do not, by themselves, establish the limits for standard Sol, Terra or Luna in Work or Codex. They do make clear that a Pro subscription is not one universal usage pool across all models and tools.

The product lineup has expanded since OpenAI introduced GPT-5.6 in June. The company describes Sol as its flagship, Terra as a balanced lower-cost model and Luna as its fastest, most affordable option. It cut Luna's API prices by 80% and Terra's by 20% on July 30th, while saying their subscription use would consume fewer credits in Codex and ChatGPT Work. In August, OpenAI also said Free users would get unlimited everyday text chats powered by Luna, subject to abuse guardrails, with separate limits on tools such as file uploads and image generation.

OpenAI has since added another model generation to the mix. On September 22nd, it launched GPT-6 Sol and Luna in ChatGPT Work and Codex and the API, saying their API prices were 50% lower than GPT-5.6 promotional pricing. The new models are not yet available in Chat, according to OpenAI. That rollout makes the distinction between model access, subscription usage and API billing more consequential: the same family name can refer to different access rules depending on where a customer uses it.

For developers and companies budgeting AI use, an unlimited subscription label cannot be translated directly into unlimited agent work or API calls. OpenAI bills API usage by tokens, while Work and Codex have their own allowances and credits. The GPT-5.6 price changes show the economic pressure behind the tiered lineup: lower-cost models widen the range of tasks that fit a subscription or an API budget, while high-compute options remain separately metered.

Nuttall's thread captures the plan-page wording and OpenAI's teaser about additions that would not count against usage. It does not establish that each GPT-5.6 model is unlimited across ChatGPT, Work and Codex, or identify which subscription features Sottiaux meant for September 30th. For now, OpenAI's published materials describe different allowances by model and product, leaving the scope of the reported label unsettled.