ChatGPT Usage Limits in 2026: What Is Still Capped and When It Resets

Jonhisking · October 2, 2026

"You've reached your limit" used to be the most common complaint about ChatGPT. In 2026 the rules changed a lot: everyday text chat no longer has a cap on any plan, but the newest models, ChatGPT Work and Codex, file uploads, images and voice still do. Here is what is limited on each plan, how the limits reset, and how to check where you stand.

Text chat: no cap since August 2026

On August 6, 2026, OpenAI removed the limit on everyday text chats for Free, Go, Plus and Pro. Free and Go users chat with GPT-5.6 Luna and can press a "Think" button for harder questions. Plus and Pro users get GPT-5.6 Sol with a thinking slider. OpenAI still applies abuse-prevention safeguards, so automated or extreme use can be throttled.

So if you only send ordinary text messages, you should no longer see a message cap.

What is still limited

Thinking levels by plan. Plus gets Medium and High thinking on GPT-5.6 Sol. Pro adds Extra High and the Pro models. On Pro 100 and Pro 200, GPT-6 Pro and GPT-5.6 Sol Pro share a weekly allowance; when it runs out, ChatGPT switches to GPT-5.6 Thinking at Medium.

GPT-6 Astra limits

GPT-6 Astra, OpenAI's flagship, is used through ChatGPT Work and Codex. OpenAI's help center gives estimated message ranges per 5-hour window:

The range is wide because one long, multi-step task uses far more of the allowance than a short one. A weekly limit applies on top of the 5-hour window. Higher Pro tiers get proportionally more: since September 29, 2026, Pro 200 is 10× Plus for new subscribers and Pro 500 is 25× Plus. See ChatGPT Pro 100 vs 200 vs 500.

How the limits reset

• 5-hour window: rolling, starting from your first request in that window, not at a fixed time of day. Plus has had the 5-hour window for Work and Codex since late August 2026. In August, OpenAI said Pro 100 and Pro 200 would not get the 5-hour window for the coming months.
• Weekly limit: a rolling seven-day allowance on Plus, Pro and Business.
• Files, images and voice: each has its own allowance.

How to check your usage

OpenAI's help center points to Settings → Usage, which shows your remaining allowance and reset times for Work and Codex. When you hit a limit, ChatGPT also tells you when you can continue.

What to do when you hit a limit

1. Wait for the reset. For the 5-hour window, usually a few hours at most.
2. Use a reset or credits. Eligible Plus and Pro accounts can apply banked resets or buy an instant reset, and some plans can continue with credits.
3. Switch model. Routine work rarely needs GPT-6 Astra; GPT-5.6 Sol or Luna use much less of the allowance.
4. Upgrade. Pro 100 is 5× Plus, Pro 200 is 10× and Pro 500 is 25×.
5. Use the API. For steady heavy use of a specific model, paying per token can be simpler. GPT-6 Astra costs $10 per million input tokens and $50 per million output tokens; see GPT-6 API pricing.

Is there a "token limit" per chat?

Every model has a context window: the maximum amount of text it can consider at once. Very long conversations or documents eventually hit it, and quality drops before that, because the model re-reads the whole conversation with every message. Starting a new chat with a short summary helps more than you might expect. See Context windows explained.

Plan or API?

If you mostly chat, the August change makes a $20 plan much better value than before. If you use Codex or GPT-6 Astra heavily, compare the plan price with what the same work costs on the API using the Subscription vs API calculator and the coding agent calculator.

Limits change often. OpenAI's GPT-6 Astra usage help page has the current numbers.