What changed: OpenAI added two new models to the GPT-6.2 family: GPT-6.2 Astra, positioned as its top-tier model for complex professional projects, and GPT-6.2 Sol, a lower-cost option for everyday work. Both models reportedly use fewer output tokens than prior versions while improving performance on coding, computer use, and multi-step agentic tasks. The models rolled out via ChatGPT Work, Codex, and the OpenAI API; Astra's standard API price was cut 20% compared to GPT-6 Astra, while Sol's pricing stayed the same as GPT-6.1 Sol.
Who is affected
Plus, Pro, Enterprise and Edu plan users on ChatGPT Work and Codex, plus developers and businesses accessing the models through the OpenAI API; the source does not mention other restrictions.
What it means for your business
For businesses using OpenAI's API or ChatGPT Work, this means a new model option for coding, document/slide generation and multi-step workflows, plus a 20% lower input/output price for Astra.
What to do
If you use the OpenAI API or ChatGPT Work/Codex, test GPT-6.2 Astra and Sol on your current coding and document workflows, and update your cost estimates to Astra's new $8 input / $0.40 cached input / $40 output per million tokens, and Sol's $2 input / $0.10 cached input / $10 output per million tokens.
When it takes effect
The source gives no date; it states the models were rolled out, but no effective date is written in the text.
Source (press reports): Webrazzi, 9to5Google, The Decoder, ShiftDelete