GPT-6.2 Astra And Sol: Three Things To Know About Work
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: GPT-6.2 Astra And Sol: Three Things To Know About Work on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get everyday essentials delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026, alongside a legal configuration called Astra for Law 6.2. Independent tests put Astra near the top of a major model index and report lower costs per benchmark task than several rivals, while finding no improvement in coding-agent performance; OpenAI rates both new models Critical for cybersecurity risk.

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026, pitching Astra for demanding projects and Sol for lower-cost everyday work. Independent testing found strong but mixed benchmark results, while OpenAI’s system card rated both models Critical for cybersecurity risk, its higher risk threshold.

OpenAI’s developer guide recommends Astra for the most demanding tasks and Sol for complex work at lower cost. The company also listed Astra for Law 6.2 as a legal model configuration. The pages available to the source did not explain how that configuration differs from standard Astra.

Artificial Analysis scored Astra at 62 and Sol at 57 on its Intelligence Index, using each model’s Max setting, in a comparison of the best AI models for your money. Claude Fable 5.2 led the family comparison with 65. Astra’s unrounded score was 62.14, just below Fable 5.2’s 62.22 at Xhigh; Astra’s individual model page ranked it third among 91 configurations. Artificial Analysis said Astra gained five points over GPT-6 Astra, while Sol gained two.

The same source reported a cost of $4.00 per index task for Astra Max and $1.01 for Sol Max, figures relevant to comparing the best AI models for your money. It listed $26.54 for Claude Fable 5.2 Max and $5.48 for Gemini 4 Carbon High on that task. These are benchmark task costs, not a guarantee of what every user’s workload will cost. Artificial Analysis also reported no improvement in its Coding Agent Index and hallucination rates of 59% for Astra and 55% for Sol at maximum effort.

OpenAI’s listed standard API prices are $8 per million input tokens and $40 per million output tokens for Astra, down 20% from GPT-6 Astra, which was listed at $10 and $50. Sol is listed at $2 per million input tokens and $10 per million output tokens. Astra for Law has the same listed rates as Astra, with promotional pricing available at least through December 31, 2026, according to OpenAI.

At a glance
announcementWhen: Released November 17, 2026; independent…
The developmentOpenAI launched GPT-6.2 Astra and Sol, and same-day independent tests found strong but uneven results alongside lower task costs and high cybersecurity risk ratings.

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

Costs, Risk and Workload Trade-offs

The release gives developers a two-tier choice: a higher-priced model aimed at difficult projects and a cheaper option for complex tasks deployed more broadly. The index’s per-task estimates suggest Sol could cost less on that particular evaluation, but actual bills depend on prompt length, processing speed, and request type.

Pricing is not as simple as the headline token rates. OpenAI says requests exceeding 272,000 input tokens incur higher rates across the whole request: input and cached input prices double, and output costs 1.5 times the short-context rate. Batch processing costs half the Standard rate, while Fast costs twice as much. Ultrafast costs three times Standard through February 28, 2027, then six times from March 1, according to the source’s account of OpenAI’s pricing page.

The risk finding matters for organizations considering models that can assist with software or technical work. OpenAI’s Critical cybersecurity rating means access to more permissive cyber capabilities now requires additional vetting, according to its system card. The release is therefore not simply a question of benchmark rank: buyers must weigh performance, costs, safeguards, and deployment requirements.

Products Worth Considering

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

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

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

As an affiliate, we earn on qualifying purchases.

How the New Models Are Positioned

OpenAI describes Astra as its most capable model for difficult projects and Sol as its choice for everyday work, including coding and powering agents at scale. Both accept text and image input and produce text. Each has a context window of 1,050,000 tokens, 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.

Artificial Analysis evaluates models across nine tests and reports scores, cost, and task time. Its overall finding was that OpenAI now trails only Anthropic’s Claude Fable 5.2 in the model-family comparison, but the lead is narrow and rankings vary by configuration. The supplied source notes that Astra led some tests and trailed on others; it identifies a lead in agentic science workflows, but the available material cuts off before providing the full test-by-test results.

OpenAI also documented an API behavior change. Starting with GPT-6.2, if a conversation history is edited mid-conversation, the API drops the model’s earlier reasoning. OpenAI warns that answer quality may decline and the model may use more tokens repeating earlier work. Requests setting reasoning effort to “none” are rejected.

Open Questions on Access and Results

The available source says OpenAI’s launch post could not be accessed, so which ChatGPT plans include the models is not established here. It is also unclear from the cited product pages how Astra for Law differs from standard Astra. Artificial Analysis had not assigned Astra for Law an Intelligence Index score, although it had tested the configuration on one legal benchmark; the supplied material does not include that benchmark’s result.

The benchmark figures reflect specific model settings and a defined set of tests, not every workplace task. The source material does not provide the complete test-by-test comparison or the full results for the legal benchmark. OpenAI’s Critical cybersecurity designation is its own assessment; the supplied material does not include an independent assessment of that rating or additional details about the vetting process.

Prices can vary with usage tier and processing mode. The source lists rate limits ranging from 5,000 requests and 1 million tokens per minute to 15,000 requests and 40 million tokens per minute across listed tiers, but a particular customer’s limits depend on its account tier. Eligibility for discounted background processing and data-retention or residency options also depends on the account and request conditions described by OpenAI.

Availability, Pricing and Further Testing

Developers can check OpenAI’s model and pricing pages for current API availability, account-specific limits, and processing options. The next practical step for organizations is to test Astra and Sol against their own workloads, since index task costs and aggregate scores do not predict every deployment’s quality or bill.

Further independent results could clarify Astra for Law’s performance and fill in the test-by-test comparison. Organizations evaluating cyber-related uses will also need to follow OpenAI’s access requirements and any updates to its system card. The stated promotional pricing period for Astra for Law runs at least through December 31, 2026; OpenAI’s listed Ultrafast pricing changes on March 1, 2027.

Key Questions

What did OpenAI release?

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026, and listed Astra for Law 6.2 as a legal model configuration.

How do Astra and Sol differ?

OpenAI positions Astra for its most demanding projects and Sol for complex work at lower cost. Their listed standard API prices are $8/$40 per million input/output tokens for Astra and $2/$10 for Sol.

How did the models perform in independent tests?

Artificial Analysis scored Astra at 62 and Sol at 57 on its Intelligence Index at Max settings. Astra narrowly trailed Claude Fable 5.2 on the reported comparison. The evaluator also reported no improvement in its Coding Agent Index.

What does the Critical cyber rating mean?

OpenAI’s system card rates both models Critical for cybersecurity. According to the source, access to more permissive cyber capabilities now requires additional vetting. The supplied material does not detail all requirements.

Are the models available in ChatGPT?

The source could not access OpenAI’s launch post and does not establish which ChatGPT plans include Astra or Sol. Check OpenAI’s current product information for availability; the article’s confirmed pricing details concern the API.

Source: ThorstenMeyerAI.com

You May Also Like

Google Gemini 1.5 Unleashes AI Revolution

Nurture your curiosity with the groundbreaking advancements of Google Gemini 1.5, setting the stage for a transformative AI revolution.

How AI Will Create More Jobs: A Step-by-Step Guide

Buckle up for a journey through the incremental steps revealing how AI will revolutionize job creation, sparking curiosity and anticipation.

AI Replacing Actors: How Technology Is Transforming the EntertAInment Industry

Wondering about the future of entertainment? Explore how AI is reshaping the industry by potentially replacing actors.

Greenland Surges In Global Coverage

Greenland’s media mentions have quadrupled, signaling increased international interest amid climate and geopolitical concerns, though the cause remains unconfirmed.