🔍 Read the full analysis: A Look At OpenAI’s GPT-6.2 Astra And Sol For Work on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, 2026, positioning them for demanding projects and lower-cost everyday work. Artificial Analysis scores Astra just behind Claude Fable 5.2 among model families, while OpenAI’s safety report rates both new models Critical for cybersecurity. Pricing and benchmark results vary by configuration, and details about ChatGPT availability were not confirmed in the source material.
OpenAI describes Astra as a model for hard projects handled from start to finish, and Sol as its option for everyday work, including coding and powering agents at scale. A third entry, GPT-6.2 Astra for Law, also appeared on the company’s pricing page as a legal model configuration. The material available to the source does not explain how that configuration differs from Astra.
Artificial Analysis’s Intelligence Index, accessed on Nov. 17, gave Astra a score of 62 at Max and Sol 57 at Max. Claude Fable 5.2 scored 65 in the rounded family-level comparison. Astra’s model page gives its unrounded score as 62.14 and places it third among 91 configurations; Fable 5.2 at Xhigh scored 62.22. Artificial Analysis said Astra gained five points over GPT-6 Astra, while Sol gained two and reached a score comparable to GPT-6 Astra at about 15% of its cost per task.
The service’s estimate for one index task was $4.00 for Astra Max and $1.01 for Sol Max, compared with $26.54 for Claude Fable 5.2 Max and $5.48 for Gemini 4 Carbon High. These figures are benchmark-task estimates, not a universal price for using a model. OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens, 20% below the previous GPT-6 Astra’s listed rates. Sol remains at $2 per million input tokens and $10 per million output tokens.
The independent results were not uniformly positive. Artificial Analysis reported no improvement in its Coding Agent Index and hallucination rates of 59% for Astra and 55% for Sol at maximum effort. Its index also showed Astra leading some tests and trailing on others. The source material does not provide enough detail to characterize every test result beyond those reported findings.
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
Performance Gains Meet Safety Trade-Offs
The release gives developers another choice between a higher-capability model and a lower-cost option, with Artificial Analysis reporting substantially lower per-task costs for Astra and Sol than for some close competitors on its benchmark. That may matter to organizations running repeated workloads, where both model quality and task expense affect deployment decisions.
But benchmark position alone does not settle whether a model is suitable for a given job. The reported coding-agent result did not improve, and the hallucination rates remain a concern for work where unsupported answers carry costs. OpenAI’s Critical cybersecurity rating also makes safety controls relevant to access decisions, not just model selection. The same release therefore combines a competitive performance showing with risk and reliability questions that customers must weigh.
Costs can also differ from the headline token rates. OpenAI’s pricing information says requests exceeding 272,000 input tokens are charged at higher rates across the whole request, while Fast and Ultrafast processing cost more than Standard. The effective bill will depend on prompt size, processing tier, request type and account usage tier.
Products Worth Considering

Sony WH-1000XM5 Wireless Noise-Canceling Over-Ear Headphones, Black, International Version

Sony WH-1000XM6 Wireless Noise-Cancelling Headphones (Black)

Samsung 27-Inch Odyssey OLED G5 QHD Gaming Monitor (G50SF)
As an affiliate, we earn on qualifying purchases.
Model Choices, Pricing and API Changes
OpenAI’s developer guide frames the choice as Astra for the most demanding tasks and Sol for complex work at lower cost. Both models accept text and images and generate 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, 2026. Fine-tuning is not supported.
The listed standard rates are $8 input and $40 output per million tokens for Astra, and $2 input and $10 output for Sol. Cached input is listed at $0.40 and $0.10, respectively. Batch processing costs half the Standard rate, Fast costs twice the rate, and Ultrafast is listed at three times the Standard rate through Feb. 28, 2027, then six times from March 1, 2027. Eligible background requests receive lower rates, subject to completion windows. Astra for Law has the same listed standard rates as Astra, with promotional pricing stated to be available at least through Dec. 31, 2026.
OpenAI also flagged an API behavior change: from GPT-6.2, editing conversation history mid-conversation drops the model’s earlier reasoning, which the company says can reduce answer quality and lead to repeated work and greater token use. The API rejects requests setting reasoning effort to “none.” OpenAI’s launch post was not accessible to the source, so the material does not establish which ChatGPT plans include the models.
Access, Legal Configuration and Risks
Several practical details remain unsettled in the source material. OpenAI’s launch post was not accessible, so ChatGPT plan availability and the timing of access for different customers could not be confirmed. The pages reviewed also did not explain what changes in Astra for Law beyond identifying it as a legal configuration. Artificial Analysis had not assigned that version an Intelligence Index score, though it had tested it on one legal benchmark; the result is not included in the available material.
The benchmark figures reflect particular configurations and test methods, and should not be treated as direct predictions of performance on every customer workload. OpenAI’s system card rates both models Critical for cybersecurity and says more permissive cyber access requires additional vetting. Astra is also described as the first model OpenAI has externally released to meet its High threshold for AI self-improvement. The source does not provide enough detail here to assess the practical likelihood or consequences of those risks. OpenAI’s safety report also lists regressions alongside improvements, but the available account does not identify all of them.
Availability and Testing to Watch
Developers can consult OpenAI’s model and pricing pages for listed rates, limits and processing options, but the available source does not establish ChatGPT rollout details. Account-specific rate limits vary by usage tier, and data-retention and residency options depend on eligibility and processing mode. Ultrafast processing is listed with US data residency only.
Further independent testing could clarify Astra for Law’s results and whether the early coding-agent and hallucination findings hold across additional tasks. Customers considering deployment will also need to check OpenAI’s current access and safety requirements, especially for cybersecurity-related use. Pricing should be reviewed against actual prompt lengths and service tiers rather than inferred from the benchmark cost-per-task figures.
Key Questions
What did OpenAI release?
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, 2026, plus a listed legal configuration called GPT-6.2 Astra for Law.
Which model performed better in the reported index?
Astra scored 62 at Max and Sol scored 57 in Artificial Analysis’s Intelligence Index. Astra’s unrounded score was 62.14, placing its configuration third among 91; Claude Fable 5.2 at Xhigh scored 62.22.
How much do the models cost?
OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens. Sol is listed at $2 per million input tokens and $10 per million output tokens. Longer prompts and faster processing tiers can raise costs.
What safety concern did OpenAI report?
OpenAI’s system card rates both models Critical for cybersecurity, its higher risk threshold, and says more permissive cyber access requires extra vetting. The source material does not provide enough detail to quantify the risks.
Are the models available in ChatGPT?
The available source did not confirm which ChatGPT plans include the new models. It reports API model and pricing information, but not the consumer product rollout.
Source: ThorstenMeyerAI.com
