AI & agents
OpenAI ships GPT-6.2 Astra and Sol together
- Published
- Source
- OpenAI Blog
Summary
OpenAI introduced GPT-6.2 Astra, its new frontier model for demanding professional work, alongside GPT-6.2 Sol for everyday tasks. Both models improve the quality of documents, slides, and code while completing the same work with fewer output tokens. By the company's own reporting, Astra scores 5.2 points higher than GPT-6 Astra on the Agents' Last Exam multistep professional benchmark at less than a third of the cost, and sets a new high on Terminal-Bench-Science 0.1 at 9 percentage points higher. API prices drop 20% versus GPT-6 Astra to $8 per million input tokens and $40 per million output tokens, with a 60% discount on cached input, and the models are live in ChatGPT Work, Codex, and the API.
Why it matters for our work
Agent-style work such as long-horizon tasks, coding, and computer use is getting better while the cost per task falls sharply. That shifts the economics of handing repetitive professional work to AI agents and could speed up team-level adoption.
Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.