Overview

• OpenAI reinstated the GPT-4o model for all paid ChatGPT users after widespread backlash over its removal in the GPT-5 launch.
• The company introduced three GPT-5 response modes—Auto, Fast, and Thinking—to give users explicit control over speed and depth of AI reasoning.
• GPT-5 Thinking now supports a 196,000-token context window and a raised rate limit of 3,000 messages per week before switching to a mini variant.
• Sam Altman pledged iterative personality updates to make GPT-5’s tone warmer and more engaging compared with the initial rollout.
• Independent firms such as SPLX and NeuralTrust have published tests revealing exploitable security and safety gaps that conflict with OpenAI’s own red-teaming assurances.