GPT-5.5 Instant free default rollout; GPT-5.6 testing; Codex automatable; Realtime-2 API; GPT Image 2; GPT-5 August reports
Key Questions
What is GPT-5.5 Instant and how is it being rolled out?
GPT-5.5 Instant has reached general availability as the default ChatGPT model, supporting context, files, and Gmail integration with improved STEM and coding performance. It achieved an 81.2% score on MMMU-Pro.
What is the current status of GPT-5.6?
GPT-5.6 is currently in testing following the GPT-5.5 Instant rollout.
What new capabilities does GPT Image 2 offer?
GPT Image 2 leads in 4K image generation and is supported by new prompt tools for creating high-quality cinematic AI images.
What features does the Realtime-2 API include?
The Realtime-2 API is now generally available with support for voice, translation, and Whisper integration, alongside GPT-4o video agents.
What is the new Codex plan and its pricing?
A new $19 Codex plan has launched with mobile support for Android and iOS, enabling automatable coding tasks.
What math breakthroughs has OpenAI reported?
An internal OpenAI reasoning model disproved an 80-year-old conjecture in discrete geometry known as the planar unit distance problem.
What insights did Yann Dubois share about AI progress?
Yann Dubois discussed post-training insights on RL scaling and reasoning, explaining why AI progress now feels more tangible.
How did ChatGPT perform on its personal finance benchmark?
ChatGPT scored 82.5 out of 100 on its own personal finance benchmark after the feature launch.
GPT-5.5 Instant GA as ChatGPT default with context/files/Gmail, stronger STEM/coding, 81.2% MMMU-Pro. GPT-5.6 testing. GPT Image 2 4K leads; Realtime-2 API GA with voice/Translate/Whisper; GPT-4o video agents. New $19 Codex plan + mobile support. August GPT-5 reports, math breakthroughs, Yann Dubois post-training insights on RL scaling/reasoning, and ChatGPT Tasks scheduled prompts. ChatGPT personal finance benchmark 82.5.