OpenAI Bundles 'Unlimited' GPT Access
OpenAI is packaging unlimited access to GPT‑5.3 and GPT‑5.4 in its ChatGPT Business and Pro tiers, placing heavy usage behind product guardrails rather than raw infrastructure guarantees. The move signals product‑level capacity management even as surface‑level limits disappear for customers. (help.openai.com)
OpenAI has started selling “unlimited” GPT access in ChatGPT while keeping the actual controls behind plan rules, model routing, and usage guardrails. (help.openai.com) In ChatGPT, GPT-5.3 Instant is now the default model for logged-in users, and OpenAI says the “Auto” setting can switch a request to GPT-5.4 Thinking when the task is more complex. GPT-5.4 Thinking is positioned for spreadsheet work, coding, slides, math, document analysis, image understanding, tool use, and web research. (help.openai.com) For ChatGPT Business, OpenAI says the plan offers “virtually unlimited messages for the base models,” with limits enforced through its Services Agreement rather than a simple message cap. The company lists banned uses including automated data extraction, account sharing, resale, and using ChatGPT to power third-party services. (help.openai.com) The pricing structure shows where the hard edges still sit. In OpenAI’s Business rate card, GPT-5.3 Instant is listed as unlimited, while GPT-5.4 Thinking costs 10 credits per message and GPT-5.4 Pro costs 50 credits per message under flexible pricing. (help.openai.com) That setup separates a ChatGPT subscription from raw computing capacity. OpenAI is selling a product experience with model access, routing, and policy controls, not promising unconstrained access to every high-cost model path. (help.openai.com; help.openai.com) OpenAI has been moving in that direction across ChatGPT this spring. On March 5, 2026, it introduced GPT-5.4 Thinking in ChatGPT and said the model can provide an upfront plan while it works, maintain context longer, and improve deep web research for highly specific queries. (openai.com) OpenAI’s release notes also show how “unlimited” and “fallback” now coexist. On March 18, 2026, the company said GPT-5.4 mini would serve as a fallback for GPT-5.4 Thinking when paid users hit rate limits, so reasoning access could continue during heavy usage. (help.openai.com) The same product split is visible in Business pricing. Standard ChatGPT Business seats now cost $25 per user each month, or $20 per user each month on annual billing in most countries, and include baseline access plus optional workspace credits for usage beyond included limits. (help.openai.com) ChatGPT Pro is pitched the same way for individuals. OpenAI’s Pro help page says the $100 plan is built for “real projects” and includes unlimited access to GPT-5, while separately describing higher allowances for advanced tools such as Deep Research and Codex. (help.openai.com) OpenAI is also retiring older model names as it does this. Its help center says GPT-4o, GPT-4.1, GPT-4.1 mini, OpenAI o4-mini, and GPT-5 were retired from ChatGPT on February 13, 2026, with GPT-4o in Custom GPTs for Business, Enterprise, and Edu remaining only until April 3, 2026. (help.openai.com) The result is a simpler pitch on the surface: pay for ChatGPT, get “unlimited” access, and let the product decide when to route, meter, or fall back. The limits did not disappear; OpenAI moved more of them into the product itself. (help.openai.com; help.openai.com)
