‘Mythos-like hacking, open to all’: Industry reacts to OpenAI’s GPT 5.5

OpenAI released GPT-5.5 and GPT-5.5 Pro, positioning them as a new class of intelligence for real work. Early developer reactions are mixed: Simon Willison found it fast and capable but noted the lack of API access and higher pricing (roughly 2x GPT-5.4). Soumitra Shukla praised its reduced need for hand-holding on longer tasks. Security firm Xbow reported GPT-5.5 cut vulnerability miss rates to 10% from 40% in GPT-5, dubbing it 'Mythos-like hacking, open to all' — though that comparison is contested since Anthropic's Mythos isn't publicly available. Ethan Mollick argues the model represents genuine progress but notes the 'jagged frontier' of AI capability persists, with gains uneven across structured vs. open-ended tasks. API access is delayed pending additional safety review, and the higher cost may keep GPT-5.4 relevant as a cheaper alternative.