OpenAI is removing text-chat rate limits for ChatGPT free and Go users starting next week, and switching the default model on those tiers to GPT-5.6 Luna, the company announced on August 6, 2026. The move arrives as ChatGPT crossed 1 billion weekly users, and it collapses one of the most visible frictions between the paid and free product: how many messages you can send before getting throttled. File uploads, image generation, and voice will keep their own separate caps.
The unlimited-text policy is paired with a new Think button for free and Go users, letting them route harder questions through higher reasoning. GPT-5.6 Luna, which replaces GPT-5.5 as the default, is the model behind the change. OpenAI's internal evaluation puts factual errors 62% lower on Luna than on GPT-5.5-Instant, the baseline it is displacing on the free tier.
Plus and Pro subscribers get a parallel upgrade. The version of GPT-5.6 Sol that runs inside ChatGPT is being retuned to be more reliable on facts, dates, numbers, and sourced answers, with tighter formatting and fewer digressions. OpenAI reports a 68% drop in factual errors on that model relative to GPT-5.5-Instant. The Sol variants that power Codex and Work are unchanged.
Key facts
- 01ChatGPT free and Go users get unlimited text chats starting next week, with limits still applying to files, images, and voice.
- 02GPT-5.6 Luna replaces GPT-5.5 as the default model for free and Go tiers, arriving this week.
- 03OpenAI says GPT-5.6 Luna has 62% fewer factual errors than GPT-5.5-Instant; GPT-5.6 Sol posts a 68% reduction.
- 04Plus and Pro users get an upgraded GPT-5.6 Sol on Thursday, plus a slider to control how much the model thinks.
- 05ChatGPT recently crossed 1 billion weekly users, the scale backdrop to the tier changes.
Paid users also gain a thinking slider that lets them dial up or down how much reasoning the model applies to a given query. The idea is that a short factual question does not need the same compute budget as a multi-step research task, and users, not the router, get to make the call. The updated GPT-5.6 Sol and the slider land Thursday for Plus and Pro.
The framing from OpenAI on Sol is unusually explicit about hallucination reduction as the headline feature rather than raw capability gains.
“The new GPT‑5.6 Sol is designed to make fewer mistakes — especially when answers depend on dates, numbers, sources, rules, or assumptions — by better using the sources it finds to answer your question.”— OpenAI, company statement via The Verge
The rollout schedule is staggered. GPT-5.6 Sol and the slider arrive first for paid tiers on Thursday. GPT-5.6 Luna reaches free and Go users this week. Unlimited text chats and the Think button follow next week. OpenAI is instructing Sol to avoid extra detail when it does not help, a direct nod to complaints that recent ChatGPT versions over-explain.
Making the free tier effectively uncapped for text is a competitive move as much as a product one. Google's Gemini has been closing the weekly-user gap, and free-tier friction has been one of the reasons casual users bounce between assistants. Removing the wall for the most common use case — plain text conversation — makes the free product a stickier default rather than a demo funnel toward the $20 tier.
The economics are the open question. Serving unlimited text at 1 billion weekly users on a newer model is not free, and OpenAI has not disclosed how it is provisioning capacity or whether Luna is meaningfully cheaper to run than GPT-5.5. The company's bet is that keeping free users on ChatGPT — rather than losing them to Gemini or others — is worth the inference bill, and that Plus and Pro remain differentiated by reasoning depth, multimodal quotas, and Codex integration.
Skeptics will note that self-reported hallucination reductions are hard to independently verify until third-party benchmarks catch up, and that "unlimited" in practice usually means "unlimited until abuse detection kicks in." Rate-limit removals also tend to surface fresh capacity strain, which OpenAI has hit before during major launches. The Think button and slider add user-facing controls that used to be hidden behind model pickers, but they also push more decisions onto users who may not know when higher reasoning is worth the latency.
The strategic read is that OpenAI is defending the top of the funnel. With Gemini narrowing the weekly-user gap and free tiers becoming the terrain where consumer AI habits are set, uncapping text chats is the cheapest lever OpenAI has to keep casual users from wandering. The interesting signal is not the unlimited chats themselves but the emphasis on factual reliability — an implicit acknowledgment that the next round of the consumer AI race is being fought on trust, not on flashier capabilities.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
