What the sources say
OpenAI’s GPT-6.1 Terra announcement describes stronger visual reasoning and computer use, with standard API prices 50% below GPT-6.1 Sol. These are the developer’s statements, not independent validation.
OpenAI’s deployment-safety report, published October 26, describes Critical cybersecurity capability treatment and related protections. A capability threshold is not a report that the model attacked anyone.
The discussion of cheaper digital workers, misuse and defensive applications includes hypothetical scenarios and editorial questions. Vendor demonstrations illustrate agent workflows; they are not demonstrations of Terra or evidence of actual attacks.
Primary sources rechecked October 30, 2026. Archival demonstrations remain background rather than new events.