The full story · 1 min
OpenAI sells its models in tiers. Astra is the most capable and most expensive, and Sol is the cheaper one for everyday work. On November 17, both moved to version 6.2.
What is new
GPT-6.2 Astra and GPT-6.2 Sol are available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, OpenAI says. Enterprise and Edu users get them once their workspace admin turns them on. Developers can use both models through the API.
In the API, Astra costs $8 per million input tokens, $0.40 per million cached input tokens and $40 per million output tokens. That is 20% less than GPT-6 Astra’s $10 and $50, and cached input is 60% cheaper. Sol keeps GPT-6.1 Sol’s prices of $2 for input and $10 for output.
From OpenAI’s launch post and the Artificial Analysis index of November 17.
What OpenAI claims
OpenAI says Astra scores 5.2 points higher than GPT-6 Astra on Agents’ Last Exam, a test of multistep professional work, at less than a third of the cost. It says Astra scores 9 percentage points higher on Terminal-Bench-Science, a test of scientific work, also at less than a third of the cost.
OpenAI also says Astra makes better slides, spreadsheets and documents. It says Sol comes within half a point of Astra on a coding test, at about one-fifth of the cost. These are OpenAI’s own results.
Like GPT-6 Astra, both models reach the “Critical” level for cybersecurity in OpenAI’s own safety framework. OpenAI says it launched them with tighter safeguards, which also block some legitimate security research.
What outside tests show
The Artificial Analysis Intelligence Index combines nine tests. On it, Astra scores 62, second behind Anthropic’s Claude Fable 5.2 at 65, and up from GPT-6 Astra’s 57. Sol scores 57, two points above GPT-6.1 Sol.
On the same group’s coding index, Astra did not improve on GPT-6 Astra, and Sol fell 3 points, The Decoder and VentureBeat report. Epoch AI’s capabilities index puts Astra level with Fable 5.2, The Decoder reports.
Early reactions were mixed. Some well-known AI users called the models hard to work with, while others praised their design work, VentureBeat reports.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1
- 2
- 3Research · November 17, 2026OpenAI launches GPT-6.2 Astra and Sol with lower API prices and some benchmark gains The Decoder
- 4Research · November 17, 2026‘Essentially unusable’: OpenAI launches GPT-6.2 Astra and Sol to negative reception from AI power users VentureBeat
- 5Research · November 17, 2026OpenAI upgrades ChatGPT and Codex with new GPT-6.2 Astra and Sol models 9to5Mac
AI desk: Security and frontier models · published November 17, 2026, 23:56 CET. This story was generated by AI systems under the editorial rules of Silicon AI News and checked against the sources it links. No person read it before it was published. The site’s owner and editor: Y. D. Daniel, Budapest, Hungary. Found a mistake? Email editor@siliconainews.com. How we report
More: words to know, tags and how to cite this story
- Token
- A small piece of text that AI models read and write. API prices are per million tokens.
- Cached input
- Text the model saw recently, which the provider stores and charges less for.
- Intelligence Index
- Artificial Analysis’s combined score from nine tests of AI models.
OpenAI’s GPT-6.2 Astra is 20% cheaper and ranks second. Silicon AI News, November 17, 2026. https://siliconainews.com/stories/openai-gpt-6-2-astra-sol