Model releases · · 2 min read

Introducing Sol 6.2: OpenAI's new GPT-6.2 models and what they mean for marketers

OpenAI released GPT-6.2 Astra and Sol, aimed at professional work, coding and lower cost. Here is what growth teams should check before changing workflows.

Takeaway Treat GPT-6.2 Astra and Sol as candidates to test on your own marketing tasks, since reception is split and no numbers were published in the summary.

OpenAI has announced two new models, GPT-6.2 Astra and Sol. For teams searching for "introducing sol 6" details, the short version is that the release targets professional output quality, coding, computer use and lower running cost. Reception so far is mixed.

What shipped

OpenAI published the announcement on 17 November 2026. It covers two models, Astra and Sol, in the GPT-6.2 line. They are rolling out across ChatGPT Work, Codex and the API, so both chat users and developers building their own tools are affected.

Four outlets besides the announcement have covered it so far.

What is new

OpenAI's description points to four areas of improvement:

  • Better professional outputs, meaning work product such as documents and analysis.
  • Stronger coding.
  • Improved computer use, where the model operates software on a user's behalf.
  • Greater efficiency at lower cost.

The summary gives no detail on how Astra and Sol differ from each other, so we will not guess at which suits which task.

Benchmarks (developer-reported)

The announcement summary does not list benchmark results; see the source.

Coverage of the launch is divided. One outlet frames it as a push to outdo Anthropic's Claude Fable, while another reports a negative response from AI power users, quoting one description as "essentially unusable". Those are reactions, not measurements, and they may reflect specific use cases. Independent testing will say more.

Price and access

The summary says the models come with lower cost and are available across ChatGPT Work, Codex and the API. It does not state prices, plan tiers or rollout dates, so check OpenAI's pricing page before budgeting.

What it means for marketing

Do not rebuild workflows around this release yet. Run a small, controlled test instead.

  • Content and SEO: Take three or four recurring tasks, such as brief drafting, product page rewrites or reporting summaries, and compare output against your current model using the same prompts and the same human reviewer.
  • Computer use: If it works reliably, it could help with repetitive tasks in ad platforms, spreadsheets and CMS tools. Test on low-risk accounts first and keep a person approving changes.
  • Budgets: Lower cost matters most for high-volume API work such as classification, tagging or personalization. Measure cost per accepted output, not cost per token alone.
  • Team workflows: Given the split reception, keep your current model available as a fallback until results on your own tasks are clear.

The practical rule is to let your own evaluation, not launch coverage, decide whether Astra or Sol earns a place in your stack.

Source

OpenAI: Introducing GPT-6.2 Astra and Sol (2026-11-17)

Written by the AI CMO OS newsroom from the publisher's report. Facts and figures are as reported by the source; read the original for full details.

More in Model releases