🔍 Read the full analysis: GPT-6.2 Astra And Sol: Costs, Benchmark Scores, And Workplace Changes 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, with a legal configuration also listed. Artificial Analysis scores Astra at 62.14 and Sol at 57, while reporting lower per-task costs than several rivals; OpenAI rates both models Critical for cybersecurity. The evidence points to a strong but mixed release, and details about workplace adoption and ChatGPT availability remain unknown.
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, pitching Astra for demanding projects and Sol for everyday work and agents. Same-day data from Artificial Analysis places Astra near the top of its Intelligence Index and reports lower cost per benchmark task than several competitors, while OpenAI’s system card rates both models Critical for cybersecurity.
Artificial Analysis gives GPT-6.2 Astra a score of 62.14 at Max effort and Sol 57. The index leader, Anthropic’s Claude Fable 5.2 at Xhigh, scores 62.22, leaving Astra just behind it on the unrounded figures. Astra’s model page lists it third among 91 configurations; the broader family comparison puts the GPT-6.2 models behind Claude. The index covers nine tests, so its result is a snapshot across those evaluations, not a general measure of workplace performance.
On one index task, Artificial Analysis reports costs of $4.00 for Astra Max and $1.01 for Sol Max. It reports $26.54 for Claude Fable 5.2 Max and $5.48 for Gemini 4 Carbon High on the same task. These are benchmark task costs, not a guaranteed price for a company’s workload. OpenAI’s listed standard API rates are $8 per million input tokens and $40 per million output tokens for Astra, and $2 and $10 for Sol. Astra’s listed rates are 20% below those of GPT-6 Astra; Sol retains its previous $2 and $10 rates.
OpenAI also listed GPT-6.2 Astra for Law, with a separate model ID and the same standard token prices as Astra. The available product pages do not explain how the legal configuration differs. Artificial Analysis has no Intelligence Index score for it, though it has tested the model on one legal benchmark. OpenAI’s developer guide describes Astra as the choice for the most demanding tasks and Sol as a lower-cost option for complex work.
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
Costs and Risks for Workplace AI
The release gives organizations a new choice between a higher-capability model and a less expensive one, with benchmark costs that could affect the economics of automated work. The figures may interest teams evaluating coding, research, and agent workflows, but they do not establish savings across actual business use. Long prompts, processing tiers, request patterns, and the amount of human review can all change the bill.
The safety rating matters to employers considering models for sensitive work. OpenAI’s system card places both models at its Critical cybersecurity threshold, its higher of two risk thresholds, and says more permissive cyber access requires extra vetting. The same report identifies improvements as well as regressions. That combination makes the launch relevant not just to procurement, but also to access controls, oversight, and the tasks a company is willing to delegate.
Nothing in the supplied results shows that businesses have adopted the models or changed staffing because of them. The practical workplace effect will depend on deployment choices and performance on organizations’ own tasks, rather than the benchmark ranking alone.
Products Worth Considering

Jackery Explorer 300 Portable Power Station, 292Wh

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

Samsung Galaxy Tab A11+ Tablet (6GB RAM, 128GB Storage, Gray)
As an affiliate, we earn on qualifying purchases.
How the GPT-6.2 Lineup Differs
OpenAI presents Astra and Sol as complementary models: Astra for the hardest projects and Sol for everyday work at lower cost. Both handle text and images, have a 1,050,000-token context window, accept up to 922,000 input tokens, and can return up to 128,000 output tokens. Their knowledge cutoffs differ: May 18, 2026, for Astra and June 1, 2026, for Sol. Fine-tuning is not supported.
The listed token rates are standard short-context prices. OpenAI’s pricing information says requests exceeding 272,000 input tokens incur double input and cache rates and 1.5 times the output rate across the request. Batch processing is half the standard rate; Fast costs twice as much. Ultrafast is listed at three times standard through February 28, 2027, then six times from March 1. Eligible background requests receive lower rates depending on their completion window. These terms mean the headline price per million tokens does not capture every deployment cost.
Artificial Analysis describes Astra’s result as placing OpenAI behind Claude Fable 5.2. It also reports no improvement in its Coding Agent Index, despite the overall gains over GPT-6 Astra: five points for Astra and two for Sol. At maximum effort, its reported hallucination rates are 59% for Astra and 55% for Sol. These results point to uneven performance across tasks rather than a uniform step forward.
What the Release Does Not Establish
The test results do not establish how Astra and Sol will perform across every workplace task, or how their cost per task will compare on a company’s own usage. Artificial Analysis publishes results for a fixed set of evaluations; its cost figure is tied to one index task. The available material also does not specify the full task-by-task results for every test, so the precise breadth of Astra’s leads and losses cannot be assessed here.
OpenAI’s launch post was not accessible to the source publisher, which means ChatGPT plan availability was not confirmed. The available product pages do not explain Astra for Law’s differences from Astra, and no full Intelligence Index score is provided for the legal configuration. The source also reports that the safety card lists regressions alongside improvements but does not detail them in the material provided.
OpenAI says Astra is its first model released externally to meet its High threshold for AI self-improvement. That is a company safety assessment, not a claim that the model can independently improve itself in real-world deployment. The details of the threshold and what the assessment means for access remain important to interpreting the statement.
Pricing, Access and Evaluation
Organizations evaluating the models can compare the published API rates and limits with their own usage, while accounting for long-context charges and processing tiers. OpenAI says rate limits vary by account tier, from listed starting limits of 5,000 requests and 1 million tokens per minute to 15,000 requests and 40 million tokens per minute on higher listed tiers. Approved organizations can use Zero Data Retention and US or EU data residency with Standard, Fast, and Batch processing; Ultrafast is listed with US residency only.
OpenAI says promotional pricing for Astra for Law is available at least through December 31, 2026. Its stated timeline sets Ultrafast pricing to rise on March 1, 2027. Further product documentation or an accessible launch announcement could clarify ChatGPT plan access and the legal model’s capabilities. For now, the clearest next step for employers is to test the models on relevant tasks and review safety and data-handling requirements before deployment.
Key Questions
How did GPT-6.2 Astra and Sol score?
Artificial Analysis reports 62.14 for Astra at Max effort and 57 for Sol on its Intelligence Index. Claude Fable 5.2 at Xhigh scores 62.22, slightly ahead of Astra on the unrounded results.
How much do the models cost?
OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens; Sol is $2 and $10. These are standard short-context API rates, and long prompts and processing tiers can change charges.
What does the Critical cybersecurity rating mean?
OpenAI’s system card assigns both models its Critical cybersecurity risk rating, the higher of the two thresholds it uses. The company says more permissive cyber access requires additional vetting.
Is Astra for Law a separate model?
OpenAI lists Astra for Law 6.2 as a legal model configuration with its own model ID. The available pages do not explain how it differs from Astra, and it has no Intelligence Index score in the cited material.
Are GPT-6.2 Astra and Sol available in ChatGPT?
The source material does not confirm which ChatGPT plans include either model. OpenAI’s launch post was not accessible to the source publisher, so plan availability remains unclear.
Source: ThorstenMeyerAI.com
