OpenAI Removes ChatGPT Text Chat Limits for Free Users

OpenAI is removing text-chat limits for ChatGPT's free and Go tiers next week, and swapping in a new default model, GPT-5.6 Luna, this week.

What is changing, and when

OpenAI announced the update on Thursday, noting that ChatGPT recently crossed 1 billion weekly users, according to TechCrunch. This week, Free and Go users get GPT-5.6 Luna as their new default model, replacing GPT-5.5. Next week, those same users lose rate limits on text-only chats and gain a "Think" button for harder questions.

The two changes land on different schedules. If your free account still hits a message cap today, that is expected — the model swap and the limit removal are separate rollouts a week apart. Limits on file uploads, images, voice, and image generation are unchanged; only text chats are affected.

Plus and Pro subscribers get a different upgrade: an improved GPT-5.6 Sol model, plus a new slider that lets them set "how much thought ChatGPT puts into an answer," according to The Verge. Both the updated Sol and the slider go live Thursday.

The accuracy claims, and why they need a caveat

OpenAI told The Verge that 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." The company also says responses will be "more direct" with "tighter formatting."

OpenAI's own internal evaluation, cited by TechCrunch, found factual errors dropped 62% for GPT-5.6 Luna and 68% for GPT-5.6 Sol compared with GPT-5.5-Instant. Those figures come from OpenAI's testing, not an independent benchmark, so treat them as a company claim rather than a verified result until third parties replicate them.

One wrinkle worth noting: the GPT-5.6 Sol shipping inside ChatGPT is a distinct build from the GPT-5.6 Sol used in Codex and Work, which is not part of this update.

Why the free-tier change matters more than the model swap

Rate limits on free-tier text chat have been the main lever pushing casual ChatGPT users toward a paid plan. Removing that limit while keeping media limits in place is a narrow move: OpenAI is giving away unrestricted text access while still charging for the parts of ChatGPT that cost the company more to serve — image generation, voice, and file processing.

That split matters for anyone deciding whether to upgrade. If your usage is mostly back-and-forth text — drafting, research, code explanations — the free tier is about to get a lot closer to Plus. If you regularly upload screenshots, documents, or use voice mode, the paid limits still apply and the gap stays.

Tier comparison after the rollout

Tier	Default model	Text chat limit	New control
Free	GPT-5.6 Luna	Removed (next week)	Think button
Go	GPT-5.6 Luna	Removed (next week)	Think button
Plus	GPT-5.6 Sol (updated)	Already unlimited	Thinking slider
Pro	GPT-5.6 Sol (updated)	Already unlimited	Thinking slider

Is a ChatGPT Plus subscription still worth it after this?

For pure text chat, the case for Plus gets weaker once the free tier drops its rate limits. The remaining differences are model quality and control, not access. Plus and Pro get the updated GPT-5.6 Sol, which OpenAI says is tuned specifically for fact-dependent tasks, plus the thinking slider to dial reasoning depth up or down per query. Free and Go get GPT-5.6 Luna and a single Think button — a coarser, on/off version of the same idea.

The paid tiers still win on anything outside plain text: unlimited or higher-volume image generation, voice mode, and file uploads keep their separate limits regardless of this update. So the upgrade decision now comes down to a narrower question than before — not "can I chat enough on free," but "do I need the more accurate model, the reasoning slider, or the media features enough to pay for them."

The scale context

OpenAI's framing ties this release to ChatGPT's growth: the product crossed 1 billion weekly users just before the announcement, per TechCrunch. Removing text limits for free users is a retention and reach move at that scale — it lowers the friction that pushes casual users to competitors offering unlimited free chat, without giving up the media features OpenAI can still charge for.

That framing also explains the staggered rollout. Shipping the new default model first, then removing limits a week later, lets OpenAI absorb the load of a much larger pool of unlimited free-tier conversations on a model it has already tuned for this update, rather than doing both at once.

What to watch next

Watch for OpenAI to publish methodology behind the 62%/68% error-reduction figures — right now they are unverified marketing numbers, not a disclosed benchmark. Also watch whether the "Think" button rollout for Free and Go actually lands on schedule next week, since the model swap and the limit removal are shipping on two separate dates rather than together.