Business Standard seats, per five hours
Sol 10–100 messages, Terra 25–200, Luna 250–2,000. OpenAI calls these “local-message estimates” and notes that cloud chats can use more of an allowance than local messages.
Reviewed guide · September 1, 2026
ChatGPT Thinking limit: reasoning is where the caps live now
With text chat unlimited on every plan, the limit people hit most is reasoning. GPT-5.6 offers Instant plus a ladder of reasoning options, and each plan gets a slice of that ladder with an allowance. This page explains the ladder, the caps and the fallback.
Reviewed: . Figures marked official come from OpenAI’s help center; everything else is labelled as observed or approximate.
The ladder
Reasoning options by plan.
From OpenAI’s GPT-5.6 help article.
| Plan | Model | Reasoning options | At the cap |
|---|---|---|---|
| Free | GPT-5.6 Luna | Think | Continues without extended reasoning until reset |
| Go | GPT-5.6 Luna | Think | Same as Free, higher allowance |
| Plus | GPT-5.6 Sol | Instant, Medium, High | Continues on another available reasoning model |
| Pro | Sol + Sol Pro | Instant through Pro | Continues on another available reasoning model |
| Business / Enterprise | Sol + Sol Pro | Instant through Pro | Continues; workspace credits can extend usage |
The numbers
What OpenAI actually publishes.
Premium seats
5x the Standard allowance and no five-hour limit.
Consumer plans
No published figure. Plus behaves in the same range as a Standard seat; Pro is 5x or 20x Plus. Anything more precise you read is an estimate.
Why a range
Reasoning cost depends on how long the model thinks, which depends on the question, the effort setting and the context length. A number of messages is, in OpenAI’s words, not a reliable measure.
Make it last
Spend reasoning where it changes the answer.
Instant first
Draft, rewrite, summarise and ask factual questions on Instant. It is unlimited.
Escalate once
Move to Medium or High for the single step that needs it, then drop back.
Specific prompts
One prompt with the goal, constraints and desired output uses less allowance than a conversation of corrections.
Fresh chat for new tasks
Very long threads make every reasoning turn more expensive.
Read the reset
The message shows exactly when the option returns.
Alternatives
What to use when reasoning is capped.
If a cap lands mid-task and you cannot wait for the reset, the practical fix is a second tool. None of these are affiliated with OpenAI; pick whichever fits the job.
Alternative
Claude
claude.aiStrong writing, analysis and long documents. Has its own five-hour and weekly limits.
AlternativeGemini
gemini.google.comBest inside Google Workspace and for Google-context tasks.
ResearchPerplexity
perplexity.aiResearch answers with web sources and citations.
AlternativeCopilot
copilot.microsoft.comUseful inside Microsoft 365 and Windows.
Multi-modelMultipleChat
multiplechat.aiSeveral models in one workspace so one provider’s cap does not end your day.
Disclosure: MultipleChat is built by the team that runs this site. It is listed as one option among several; nothing on this page is paid placement.
Official sources
Where to verify the current rule.
Checked September 1, 2026. OpenAI changes plan rules without notice; these pages and your own in-app counter are authoritative.
OpenAI Help: GPT-5.6 in ChatGPT
Which reasoning options each plan gets, unlimited text on Free and Go, and what happens at a reasoning limit.
OpenAI Help: ChatGPT Business models and limits
The only official per-window message estimates OpenAI publishes (Standard and Premium seats).
ChatGPT pricing
Current plan names, prices and what each tier includes.
FAQ
Thinking and reasoning limit questions answered.
What is the ChatGPT Thinking limit?
A metered allowance on the extended-reasoning options of GPT-5.6. Free and Go get the Think option only; Plus gets Medium and High; Pro, Business and Enterprise get everything up to Pro reasoning. When you reach it, ChatGPT continues on another available reasoning model and shows a reset time.
How many Thinking messages do I get?
OpenAI publishes consumer numbers only indirectly. Its Business page gives 10–100 Sol messages per five hours for Standard seats as “local-message estimates”; Plus is comparable. Treat any exact figure you see elsewhere as an estimate.
Does the limit block my chat?
No. The chat continues on a lower reasoning setting or the Instant model. Only the higher reasoning option is paused until the reset.
What is the difference between Medium, High, Extra High and Pro?
How long the model reasons before answering. Higher settings cost more allowance per message and are gated by plan: Medium and High on Plus, Extra High and Pro on Pro, Business and Enterprise.
Does the Think button on Free use the same limit?
It is Free and Go’s only reasoning option, on GPT-5.6 Luna, and it has its own limit. Sol reasoning options are paid-tier features.
How do I make reasoning last?
Use Instant for drafts and questions, escalate for the hard step only, write one specific prompt instead of several vague ones, and keep an eye on the reset time.