Quick answer
OpenAI announced two related but separately timed changes for ChatGPT Free and Go:
- GPT-5.6 Luna becomes the default model during the rollout week of August 6, 2026.
- Unlimited text chats and a new Think button begin the following week, subject to abuse guardrails.
The word “unlimited” applies to text chats in this announcement. OpenAI explicitly says limits still apply to file uploads, images, and other tools. Rollout timing and account state can also differ, so the signed-in ChatGPT interface is the final access check.
Use the GPT-5.6 family page for the broader Sol, Terra, Luna, API-pricing, and Flowith boundaries.
What Free and Go users actually receive
| Access question | What OpenAI announced | What not to assume |
|---|---|---|
| Default model | GPT-5.6 Luna becomes the Free and Go default during the rollout week | Every account changes at the same moment |
| Text chat | Unlimited text chats start the following week, subject to abuse guardrails | Unlimited uploads, images, or every tool |
| Deeper reasoning | A Think button gives Luna more time for harder questions | A permanent entitlement on every product surface |
| Product scope | The change applies to ChatGPT Free and Go | The same access inside Work, Codex, the API, or Flowith |
This distinction matters when troubleshooting. A user can have Luna as the default before the unlimited-text and Think controls finish rolling out. A visible Luna label therefore does not prove that every second-stage feature is already active.
Default Luna versus the Think button
Use the normal Luna response for a direct question, quick rewrite, routine explanation, or short planning step. Use Think when the task needs more deliberate reasoning, such as:
- comparing several constraints;
- checking a multi-step plan for omissions;
- explaining a difficult concept from more than one angle;
- reviewing assumptions before a decision;
- working through a harder quantitative or logical problem.
Think gives the model more time; it does not make the answer automatically correct. Verify dates, numbers, citations, rules, and consequential recommendations against suitable sources.
Unlimited text is not unlimited product access
The announcement preserves three boundaries:
- Abuse guardrails still apply. Unlimited text is not permission to evade safety or platform controls.
- Tools retain limits. Uploads, images, and other tools can have their own allowances.
- Products remain separate. ChatGPT plan access does not establish API credits, Codex access, Work access, or a Flowith model mapping.
If a workflow depends on a file, image, connector, or another tool, check that control directly before starting a long session. Do not plan a batch around the text-chat announcement alone.
How to verify your live access
- Sign in to the intended Free or Go account.
- Start a new ChatGPT conversation and inspect the current model or product label.
- Look for the Think control in the composer or response controls.
- Test a short text-only prompt before relying on a long conversation.
- Check uploads, images, and other tools independently because their limits remain separate.
- If a control is missing, review the official announcement and current help surfaces before assuming an account problem.
Do not use screenshots from another plan, region, account, or rollout moment as proof of your own entitlement.
A practical Free and Go workflow
Start with a compact request that states the goal, relevant context, and desired output. If the first response is too shallow for the decision, use Think and ask the model to identify assumptions, uncertainties, and checks. Keep source verification outside the model response for high-stakes facts.
For example:
Compare these three options against my budget, deadline, and privacy constraint. State assumptions, show the tradeoffs, and list what I must verify before acting.
That prompt uses deeper reasoning for a bounded decision without implying that the model, plan, or tool is unlimited in every sense.
Frequently asked questions
Is GPT-5.6 Luna the default for ChatGPT Free and Go?
OpenAI announced that Luna would become the default for Free and Go users during the week of August 6, 2026. Because this was a rollout, check the signed-in ChatGPT model and controls rather than assuming every account changed at the same moment.
Do Free and Go users get unlimited GPT-5.6 Luna text chats?
OpenAI announced unlimited text chats beginning the week after August 6, subject to abuse guardrails. This does not remove separate limits for file uploads, images, and other tools.
What does the Think button do for GPT-5.6 Luna?
OpenAI describes Think as a way for Free and Go users to give Luna more time for questions that need deeper reasoning. It is not evidence that every tool, quota, or product surface becomes unlimited.
Does this announcement prove GPT-5.6 Luna is available in Flowith?
No. The announcement describes ChatGPT Free and Go. Check the live Flowith model selector and account entitlement separately.
Related GPT-5.6 guides
- GPT-5.6 Sol thought slider for Plus and Pro
- ChatGPT Sol versus Work and Codex version boundary
- GPT-5.6 family, pricing, and availability
Official source
Source check: August 8, 2026. Access, rollout timing, plan names, guardrails, and tool limits can change; verify the signed-in ChatGPT experience before relying on a feature.