🔍 Read the full analysis: A Workplace Guide To OpenAI’s GPT-6.2 Astra And Sol on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI released GPT-6.2 Astra for demanding projects and GPT-6.2 Sol for lower-cost everyday work on Nov. 17, 2026; it also listed a legal configuration of Astra. Independent testing places Astra near the top of a leading model index and reports lower cost per task than several rivals, but finds no improvement in its Coding Agent Index. OpenAI rates both models at its Critical cybersecurity threshold, and key workplace details, including ChatGPT plan availability, remain unclear.
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, 2026, positioning Astra for demanding projects and Sol for lower-cost everyday work, including coding and agents. Independent testing puts Astra close to the top of a widely followed model index, while OpenAI’s same-day safety report rates both models at its Critical cybersecurity threshold—a combination of performance gains, workplace costs and risks employers will need to weigh.
OpenAI describes Astra as its model for “our hardest projects, from start to finish” and Sol as its “go-to model for everyday work, from coding to powering agents at scale.” The company also listed GPT-6.2 Astra for Law, a legal configuration of Astra. The materials available to the source did not explain how that version differs from the standard model.
Both models accept text and images and produce text. Each has a 1,050,000-token context window, with up to 922,000 input tokens and 128,000 output tokens. Astra’s stated knowledge cutoff is May 18, 2026; Sol’s is June 1. OpenAI says fine-tuning is not supported. Its developer guide flags a behavior change: if conversation history is edited mid-conversation, the API drops earlier reasoning, which can lower answer quality and lead the model to repeat work using more tokens. Requests that set reasoning effort to “none” are rejected.
OpenAI lists standard API rates of $8 per million input tokens and $40 per million output tokens for Astra, and $2 and $10, respectively, for Sol. Astra’s input and output rates are 20% below those for the previous GPT-6 Astra; Sol retains its listed $2/$10 rates. Cached input costs less. Pricing can change substantially with usage: requests exceeding 272,000 input tokens are charged at higher rates, while Batch is half the standard rate and Fast is twice the standard rate. Ultrafast is listed at three times standard through Feb. 28, 2027, then six times from March 1.
GPT-6.2 Astra, Sol and Astra for Law
Three models, three jobs
What launched, what it costs and how it scores in independent tests. Astra and Sol at their Max setting; the better value in each row is outlined.
- Three releases, one dayOn November 17, OpenAI released GPT-6.2 Astra for the hardest projects and GPT-6.2 Sol for everyday work. Astra for Law 6.2 is a legal configuration at Astra’s price.
- The independent indexAstra scores 62, second among model families behind Claude Fable 5.2 at 65. Sol scores 57. One index task costs $4.00 and $1.01.
- Speed and a legal testAstra needs 10.4 minutes per task, Sol 7.9. On the Harvey LAB-AA legal test, Astra for Law passes 20.8% of tasks.
- OpenAI’s own safety ratingBoth models meet OpenAI’s Critical threshold for cybersecurity. Higher-risk cyber help now needs extra vetting.
Source: Artificial Analysis, model pages for GPT-6.2 Astra, GPT-6.2 Sol and Astra for Law 6.2, accessed Nov 17, 2026 (Intelligence Index, cost and time per task, Harvey LAB-AA v1.1); OpenAI, API pricing page and GPT-6.2 System Card, Nov 17, 2026. Prices per million input/output tokens. Redrawn by Thorsten Meyer AI.
Graphic: Powered by Thorsten Meyer AI
What the Results Mean at Work
For organizations choosing models across a workforce, the release presents a clear capability-versus-cost choice: Astra is the higher-performing option in the reported index, while Sol costs less per task in the same evaluation. Artificial Analysis puts the cost of one index task at $4 for Astra Max and $1.01 for Sol Max. Its reported comparisons are $26.54 for Claude Fable 5.2 Max and $5.48 for Gemini 4 Carbon High. These figures describe that index’s task and configurations, not a guaranteed price for a company’s own workload.
The benchmark picture is not uniformly positive. Artificial Analysis reports no improvement in the Coding Agent Index and gives hallucination rates of 59% for Astra and 55% for Sol at maximum effort. Those results do not by themselves predict performance in a particular workplace, but they argue against treating a high overall index score as proof that a model can safely or reliably handle every coding or research task without review.
Safety also affects deployment decisions. OpenAI’s system card rates both models Critical for cybersecurity, the higher of the two risk thresholds it uses, and says more permissive cyber access requires extra vetting. The report also identifies regressions alongside improvements. Employers therefore need to evaluate access controls and human oversight as well as price and benchmark scores, particularly before using models for sensitive work or giving agents broad permissions.
Products Worth Considering

Bose Smart Ultra Dolby Atmos Soundbar with Voice Control, HDMI eARC, Black

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptic Feedback, Graphite

Samsung 27-Inch Odyssey G5 QHD Gaming Monitor (G50F)
As an affiliate, we earn on qualifying purchases.
How OpenAI Positions the Models
OpenAI’s developer guide frames Astra and Sol as options for different workloads: Astra for the most demanding tasks and Sol for complex work at lower cost. Artificial Analysis, which evaluates models across nine tests, reported on launch day that OpenAI now trails only Anthropic’s Claude Fable 5.2 in its Intelligence Index. Its rounded summary scores are 62 for Astra Max and 57 for Sol Max; Fable 5.2 leads with 65.
The ranking depends on configuration and precision. Astra Max’s unrounded score is 62.14, compared with 62.22 for Fable 5.2 at Xhigh. Artificial Analysis lists Astra at rank 3 of 91 on its model page, even though rounded family-level summaries can make the gap appear narrower. The evaluator reports Astra improved by five points over GPT-6 Astra and Sol by two points; it says Sol’s score is on par with GPT-6 Astra at about 15% of the cost per task.
OpenAI’s safety report says Astra is the first model the company has released externally that meets its High threshold for AI self-improvement. That assessment is separate from its Critical cybersecurity rating. The source article also disclosed that it was researched and drafted with AI assistance, including Anthropic’s Claude, and that neither OpenAI, Anthropic nor Artificial Analysis reviewed or sponsored it. Its launch-day findings are attributed to the companies’ materials and Artificial Analysis, rather than presented as a sponsored comparison.
What Employers Still Need to Verify
Several practical details were not established in the materials summarized on launch day. The source could not access OpenAI’s launch post, so it could not confirm which ChatGPT plans include Astra or Sol, or whether consumer and business availability differs. The published API information does not answer how the models perform on specific company tasks, nor whether the reported benchmark results will hold across different prompts, workflows and review processes.
The legal configuration is also not fully described: OpenAI lists Astra for Law 6.2 and its model ID, but the available pages do not explain its differences from Astra. Artificial Analysis had not assigned it an Intelligence Index score, though it had tested it on one legal benchmark. The source material does not provide that benchmark’s result, so no conclusion about legal performance can be drawn here.
OpenAI’s cybersecurity and self-improvement ratings are its own risk assessments; the summary does not include enough detail to independently assess the methods or the specific regressions it reports. Actual costs will depend on usage tier, prompt length, processing speed and background settings. Organizations should check current pricing and eligibility rather than assume the headline per-token rates apply to every request.
Availability, Pricing and Testing
Organizations considering deployment can check OpenAI’s current model pages, API pricing and account-specific rate limits. Listed limits vary by usage tier, from 5,000 requests and 1 million tokens per minute to 15,000 requests and 40 million tokens per minute on the tiers described. OpenAI says approved organizations can use Zero Data Retention and US or EU data residency with Standard, Fast and Batch processing; Ultrafast is listed with US data residency only.
Teams should test the models against their own tasks before changing workflows, measuring quality, latency, token consumption and the need for human review. They should also account for the edited-history behavior in the API and review OpenAI’s access requirements for cybersecurity-related use. For legal workflows, the available information does not establish how Astra for Law differs or what safeguards it adds.
OpenAI says promotional pricing for Astra for Law is available at least through Dec. 31, 2026; the source materials do not specify what pricing follows. Further independent results, clearer product documentation and confirmation of ChatGPT plan availability would help clarify the models’ real workplace reach. Until then, the launch-day evidence supports a differentiated offering with competitive benchmark performance and lower reported task costs, not a blanket recommendation for every organization.
Key Questions
What is the difference between GPT-6.2 Astra and Sol?
OpenAI positions Astra for its most demanding work and Sol for complex everyday tasks at lower cost. Both accept text and images, have the same stated context-window limits and produce text.
How much do the models cost through the API?
Standard listed rates are $8 per million input tokens and $40 per million output tokens for Astra, and $2 and $10 for Sol. Long prompts, speed tiers and background processing can change the bill.
Which model performed better in independent testing?
Artificial Analysis reported a score of 62 for Astra Max and 57 for Sol Max, compared with 65 for Claude Fable 5.2. Astra’s unrounded score was 62.14, slightly below Fable’s 62.22 in the cited configurations; results vary by test.
Are Astra and Sol available in ChatGPT?
The source material did not confirm which ChatGPT plans include the models. OpenAI’s API pricing and model pages describe developer access, but do not settle ChatGPT plan availability.
What does OpenAI’s Critical cybersecurity rating mean?
OpenAI’s system card places both models at its Critical cybersecurity threshold, the higher of two risk thresholds it uses. The source says more permissive cyber access requires extra vetting; the rating is OpenAI’s assessment, not an independent certification.
Source: ThorstenMeyerAI.com
