OpenAI has made important updates to ChatGPT in the past week: GPT-4o returns to the Legacy model, GPT-5 adds multi-modal options, Plus and Team users get higher message limits, and GPT-5 is becoming more "warm". These adjustments significantly enhance user experience and flexibility.
1. Legacy model return
- GPT-4o is a return to the Legacy model and is provided by default for paying users.
- Users can turn on "Show additional models" in the settings and select o3, GPT-4.1, and GPT-5 Thinking mini.
- Pro users have additional access to GPT-4.5.
2. New model of GPT-5 model
- GPT-5 introduces three modes: Auto, Fast, and Thinking.
- Auto: Automatically switches between speed and depth.
- Fast: Quick response mode.
- Thinking: Deep reasoning mode.
- GPT-5 Thinking supports longer contexts, enhancing its ability to handle complex tasks.
3. Message limit increase
- Plus and Team users can use 3,000 messages per week in GPT-5 Thinking mode.
- After exceeding the amount, it automatically goes to GPT-5 Thinking mini, ensuring continuous use.
- GPT-5 is fully open to Enterprise and Edu users without strict restrictions.
4. GPT-5 personality optimization
- OpenAI said that GPT-5 will gradually have a more "warm" personality.
- Unlike GPT-4o's style, GPT-5 increases affinity while maintaining logical clarity.
- OpenAI promises to transparently inform users of future model updates and offlines in advance.
Frequently Asked Questions
Q: How do I choose the three modes of GPT-5?
A: Use Fast when you need quick results, Thinking when you need deep thinking, and Auto when you are unsure.
Q: Are there still restrictions for Plus users?
A: Yes. Plus users are limited to 160 GPT-5 messages in 3 hours, but there is an additional weekly cap and mini mode support.
Q: What distinguishes GPT-4o from GPT-5?
A: GPT-4o returns as a Legacy model, leaning towards multimodality and warmer interactions; GPT-5 offers flexible mode switching and enhanced reasoning capabilities.
Q: Is there a cap on enterprise users?
A: Enterprise and Edu users can use GPT-5 unlimitedly, subject to usage policies.
