Astra And Sol From OpenAI: GPT-6.2 Costs, Scores, And Work
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Astra And Sol From OpenAI: GPT-6.2 Costs, Scores, And Work on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, 2026, alongside a legal configuration called Astra for Law. Artificial Analysis scores Astra at 62 and Sol at 57, while reporting lower cost per task than several rivals; OpenAI’s system card rates both models Critical for cybersecurity. Some test results were mixed, and details including Astra for Law’s differences from Astra remain unclear.

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, pitching Astra for demanding projects and Sol for lower-cost everyday work, while independent testing found strong but uneven results. Artificial Analysis gave Astra a score of 62 and Sol a score of 57 on its Intelligence Index; OpenAI’s same-day safety report rated both models Critical for cybersecurity, its higher risk threshold.

OpenAI describes Astra as its model for the most demanding tasks and Sol as a choice for everyday work, including coding and powering agents at scale. The company also listed Astra for Law 6.2, a legal configuration with the model ID gpt-6.2-astra-law. The available product pages do not explain how that configuration differs from Astra.

Artificial Analysis reported Astra at 62 at its Max setting and Sol at 57. Claude Fable 5.2 leads the index’s model-family comparison at 65, according to the source’s rounded figures. The ranking is close: Artificial Analysis lists Astra’s unrounded score as 62.14 and Fable 5.2’s Xhigh score as 62.22, placing Astra (Max) third among 91 configurations on its model page.

The same index task was priced at $4.00 with Astra (Max) and $1.01 with Sol (Max), compared with $26.54 for Claude Fable 5.2 (Max) and $5.48 for Gemini 4 Carbon (High). These are Artificial Analysis cost-per-task estimates, not OpenAI’s token prices. OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens, 20% below the previous GPT-6 Astra rates; Sol is listed at $2 and $10. Cached input is priced at $0.40 per million tokens for Astra and $0.10 for Sol.

At a glance
announcementWhen: Released Nov. 17, 2026; benchmark and s…
The developmentOpenAI released GPT-6.2 Astra and Sol on Nov. 17, with same-day benchmark results and a safety report showing both models meet its Critical cybersecurity risk threshold.

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.

AstraSolFor Law Price in/out$8/$40$2/$10$8/$40 Index score6257– Cost per task$4.00$1.01– Minutes per task10.47.9– Legal test15.0%11.1%20.8% Both rated Critical for cyber risk
  1. 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.
  2. 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.
  3. 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.
  4. 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

Lower Costs, Higher Cyber Risk

The release gives developers a new choice between a higher-capability model and a lower-priced option, with independent task-cost estimates suggesting that some workloads may cost less than comparable runs on rival systems. That comparison depends on the benchmark’s task and model settings; it does not establish that every real-world workload will have the same price or performance.

The safety rating is a significant qualification. OpenAI’s system card places both models at its Critical cybersecurity threshold, and says more permissive cyber access now requires additional vetting. The report also identifies regressions as well as improvements, so the launch is not simply a story of better scores or lower costs. Artificial Analysis reports no improvement in its Coding Agent Index and hallucination rates of 59% for Astra and 55% for Sol at maximum effort.

For buyers and developers, the practical question is whether the new models’ performance, pricing, and access conditions fit a particular use case. The benchmark and safety findings offer information for that decision, but neither guarantees performance or cost in an individual deployment.

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

As an affiliate, we earn on qualifying purchases.

Model Choices and API Changes

Astra and Sol both accept text and images and produce text. OpenAI lists a 1,050,000-token context window for each, with up to 922,000 input tokens and 128,000 output tokens. Their stated knowledge cutoffs differ: May 18, 2026, for Astra and June 1, 2026, for Sol. OpenAI says fine-tuning is not supported.

List prices are only part of the API cost. OpenAI says requests exceeding 272,000 input tokens are billed at double the input and cached-input rates and 1.5 times the output rate for the whole request. Batch processing costs half the Standard rate, while Fast costs twice Standard. Ultrafast is listed at three times Standard through Feb. 28, 2027, then six times Standard from March 1. Eligible background requests receive a 30% discount for a two-hour completion window or a 40% discount for an eight-hour window.

OpenAI also documents an API behavior change: from GPT-6.2, editing conversation history mid-conversation removes the model’s earlier reasoning. The company warns that answer quality can decline and the model may use more tokens to repeat earlier work. Requests that set reasoning effort to none are rejected. The source article says OpenAI’s launch post was inaccessible, so it could not establish which ChatGPT plans include the models.

Questions About Access and Testing

The available material does not establish which ChatGPT subscription plans include Astra or Sol. It also does not explain how Astra for Law differs from standard Astra, and Artificial Analysis had not published a full Intelligence Index score for the legal configuration. Its testing of that configuration was limited to one legal benchmark in the supplied source.

The benchmark figures reflect specific evaluation settings and should not be treated as a guarantee of results in deployed applications. The source reports hallucination rates of 59% for Astra and 55% for Sol at maximum effort but does not provide further methodological detail in the supplied material. The cost-per-task figures are likewise tied to the evaluator’s test, not a universal estimate. OpenAI’s safety report is described as listing both improvements and regressions, but the supplied text does not detail each one.

OpenAI’s launch post was not accessible to the source writer. The reported prices, limits and availability details were drawn from OpenAI pricing and model pages accessed on Nov. 17, 2026; account rate limits vary by usage tier. The source also discloses that its research and drafting used AI assistance, including Anthropic’s Claude, whose models appear in the comparisons, and says neither OpenAI, Anthropic nor Artificial Analysis reviewed or sponsored the article.

Pricing and Access Details Ahead

Developers can compare the models using OpenAI’s published API prices, context limits and usage-tier rate limits, while checking the applicable processing option and prompt length before estimating a bill. For approved organizations, OpenAI lists Zero Data Retention support and US and EU data residency with Standard, Fast and Batch processing; Ultrafast is listed with US data residency only.

OpenAI says promotional pricing for Astra for Law is available at least through Dec. 31, 2026. Further information about the legal configuration, ChatGPT plan access and the models’ performance across additional independent evaluations would help clarify how broadly the release applies. Until those details emerge, the confirmed picture is a high-scoring release with lower listed prices than its predecessor, mixed evaluation results and heightened cybersecurity controls.

Key Questions

What did OpenAI release?

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, 2026, and listed a legal configuration called Astra for Law 6.2 on the same day.

How did Astra and Sol score?

Artificial Analysis reported scores of 62 for Astra (Max) and 57 for Sol (Max) on its Intelligence Index. The source’s unrounded Astra score was 62.14, narrowly below Claude Fable 5.2 at 62.22 in the cited comparison.

How much do the models cost?

OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens, and Sol at $2 and $10. Actual charges can vary with prompt length and processing tier; requests over 272,000 input tokens have higher rates under OpenAI’s listed pricing rules.

What does the Critical cybersecurity rating mean?

OpenAI’s system card says both models meet its Critical threshold for cybersecurity, its higher of two risk thresholds. The supplied source says more permissive cyber access requires additional vetting; it does not provide further detail about specific access decisions.

Is Astra for Law different from Astra?

OpenAI lists Astra for Law 6.2 as a legal model configuration, but the available pages cited in the source do not explain how it differs from standard Astra. Artificial Analysis had not assigned it an Intelligence Index score.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

Saratech Expands Siemens PLM Reach With Acquisition Of Geometric Solutions

Saratech acquired Michigan-based Geometric Solutions, taking over Siemens PLM sales, implementation, support and training for its customers.

From AI Tools To Video Editing: 6 Creator Laptops To Consider In 2027

A source roundup compares six laptops for creator workflows, from a 2-in-1 HP to large-screen NIMO models. Performance and pricing claims need verification.

From Arkéa’s Own Account: What We Know About The €100 Million Hack

Crédit Mutuel Arkéa says attackers used the bank’s own account through a service provider. Customer accounts were not used; the investigation is ongoing.

Why AI Camera Lenses Are Essential For 2026 Photography

Exploring the critical role of AI-enhanced camera lenses in shaping professional and amateur photography in 2026.