What are the new rate limits for plus ChatGPT subscribers on all models?
- User
What are the new rate limits for plus ChatGPT subscribers on all models?
Assistant
Okay, here's the breakdown of the new rate limits for ChatGPT Plus subscribers, focusing on the most recent updates with the introduction of GPT-4o:
-
GPT-4o:
- Plus subscribers get up to 80 messages every 3 hours on GPT-4o.
- This is a significant increase (essentially double) compared to the previous GPT-4 limit.
-
GPT-4:
- When GPT-4o is unavailable (due to hitting its limit or during peak times), or if you specifically select GPT-4, the usage counts towards the same shared limit as GPT-4o.
- So, the combined usage cap for GPT-4 and GPT-4o is 80 messages every 3 hours. You don't get separate limits for each.
-
GPT-3.5:
- For Plus subscribers, the limits on GPT-3.5 are much higher and generally not a practical constraint for typical conversational use. OpenAI doesn't usually state a precise number, but it's significantly higher than the GPT-4/4o cap, allowing for much more frequent interaction.
Key Points to Remember:
- Shared Cap: The most important limit for Plus users is the combined 80 messages / 3 hours cap for GPT-4 and GPT-4o.
- Dynamic Adjustments: OpenAI notes that limits might be adjusted dynamically based on current system load and demand. While 80/3hrs is the stated target, it could potentially vary slightly in practice during extremely high usage periods.
- Interface Indication: ChatGPT usually tells you when you are approaching or have hit your limit for the GPT-4/4o models and when the limit will reset.
- API vs. ChatGPT: These limits are specifically for using the ChatGPT interface (web or app). API usage has entirely separate rate limits based on tiers and billing.
In summary, the main change is the increased message cap to 80 messages every 3 hours for Plus users when using the advanced GPT-4 and GPT-4o models, while GPT-3.5 remains largely unconstrained for typical use.
-
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.
Identical first prompt
- gpt-4o-search-previewCompare side by side →
gemini-2.5-flash-preview-04-17-groundingCompare side by side →
gemini-2.5-pro-exp-03-25-wo-searchCompare side by side →
sonarCompare side by side →
sonar-pro-highCompare side by side →
sonarCompare side by side →
gemini-2.0-flash-groundingCompare side by side →
sonarCompare side by side →
gemini-2.0-flash-groundingCompare side by side →