OpenAI CEO Sam Altman said users can now select between “Auto,” “Fast,” and “Thinking” modes for GPT-5 in ChatGPT, offering more control over model performance.

Paid ChatGPT users now have access to the 4o model by default, and can use a new toggle to access additional models such as o3, 4.1, and GPT-5 Thinking mini.

GPT-5 Thinking mode has a weekly rate limit of 3,000 messages, with continued access via GPT-5 Thinking mini after the cap is reached.

The context limit for GPT-5 Thinking is set at 196,000 tokens.

Altman said 4.5 remains available only to Pro users due to high GPU costs.

He also said OpenAI is working on an update to GPT-5’s personality, and plans to give users more customization options in the future.

OpenAI would like to allow Plus users a limited number of Pro queries in the future, but does not currently have the compute capacity to do so.

🔗 Source: Sam Altman

🧠 Food for thought

1️⃣ AI personality optimization reflects growing user experience priorities

Altman’s focus on refining GPT-5’s personality to be “warmer but not as annoying” aligns with research showing AI personality significantly impacts user satisfaction and trust.

Studies demonstrate that users desire not just accurate answers but relatable interactions, with well-crafted AI personalities making exchanges more engaging and building stronger rapport with users1.

The challenge lies in balancing professionalism with friendliness, which OpenAI appears to be addressing by moving away from what some users found annoying in GPT-4 while maintaining warmth.

This personality refinement reflects a broader industry recognition that successful AI deployment depends on emotional connection, not just technical capability.

2️⃣ Per-user customization addresses rising personalization expectations

Altman’s acknowledgment that OpenAI “really just needs to get to a world with more per-user customization of model personality” responds to documented consumer demands for tailored AI experiences.

Recent OpenAI developments