GPT-6.2 Astra And Sol: What OpenAI’s New Models Cost, Score And Change For Work
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: GPT-6.2 Astra And Sol: What OpenAI’s New Models Cost, Score And Change For Work on ThorstenMeyerAI.com

Before you orderOffer from Amazon

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
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 Nov. 17, 2026, with Astra positioned for demanding projects and Sol for lower-cost everyday work. Artificial Analysis scores Astra at 62 and Sol at 57; its task-cost estimates are lower than those for the listed competitors, but it reports no improvement in its Coding Agent Index. OpenAI rates both models Critical for cybersecurity.

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17, 2026, adding a high-capability model aimed at demanding projects and a lower-cost option for everyday work and agents. Independent tests put Astra near the top of Artificial Analysis’s Intelligence Index, while OpenAI’s safety report rates both models Critical for cybersecurity and lists performance regressions as well as gains.

Artificial Analysis gives GPT-6.2 Astra a score of 62 at its Max setting and Sol a score of 57. Its index leader, Anthropic’s Claude Fable 5.2, scores 65 in the headline comparison. However, the ranking depends on configuration: Astra’s model page lists it third among 91 configurations, with an unrounded score of 62.14, just below Fable 5.2 Xhigh at 62.22. The release is therefore competitive, but it is not the index leader.

The benchmark provider estimates that one index task costs $4.00 with Astra Max and $1.01 with Sol Max. Its comparable figures are $26.54 for Claude Fable 5.2 Max and $5.48 for Google Gemini 4 Carbon High. These are Artificial Analysis task-cost estimates, not a guarantee of what every customer will pay; actual bills depend on usage and API settings. OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for GPT-6 Astra. Sol is listed at $2 input and $10 output per million tokens.

Both models accept text and images, have a 1,050,000-token context window, and can return up to 128,000 output tokens. OpenAI also listed GPT-6.2 Astra for Law, model ID gpt-6.2-astra-law, under legal configurations. Its standard listed rates match Astra’s, with promotional pricing available at least through Dec. 31, 2026, according to OpenAI. The pages available to the source report did not explain how the legal configuration differs from Astra.

At a glance
announcementWhen: Announced November 17, 2026; benchmark…
The developmentOpenAI launched GPT-6.2 Astra and Sol, alongside a legal configuration, as independent benchmarks and OpenAI’s safety report offered a mixed early assessment.

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 Task Costs, Mixed Results

The launch gives developers another choice between capability and cost: Astra is aimed at the most demanding work, while Sol offers a less expensive route for complex workloads. Artificial Analysis’s task estimates suggest that the models could reduce the cost of its tested tasks compared with the listed rivals. Those figures may matter to teams running agents or repeated model calls, where usage can add up.

The score alone does not establish that either model will perform better for a particular workplace. Artificial Analysis reports no improvement in its Coding Agent Index and gives hallucination rates of 59% for Astra and 55% for Sol at max effort. Those results complicate the productivity case: businesses evaluating the models need to test them on their own coding, research, and operational tasks, not infer reliability from a headline index score.

Safety findings also affect deployment decisions. OpenAI’s system card places both models at its Critical cybersecurity threshold, its higher of two risk thresholds. OpenAI says more permissive cyber access now requires extra vetting. It also says Astra is the first model it has released externally to meet its High threshold for AI self-improvement. These are OpenAI’s assessments and controls; the supplied material does not provide an independent review of those ratings.

Products Worth Considering

Bose QuietComfort Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Eucalyptus Green

Bose QuietComfort Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Eucalyptus Green

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

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

Kindle Paperwhite Signature Edition 32GB (2024 Model) - Metallic Black

Kindle Paperwhite Signature Edition 32GB (2024 Model) – Metallic Black

As an affiliate, we earn on qualifying purchases.

How OpenAI Positions the Models

OpenAI’s developer guide summarizes the intended split this way: “Use GPT-6.2 Astra for the most demanding tasks or GPT-6.2 Sol for complex work at lower cost.” The company describes Astra as suited to hard projects from start to finish and Sol as its go-to for everyday work, coding, and agents at scale. Those descriptions are product positioning, not a claim that independent tests show Astra leading every category.

Pricing has several qualifications beyond the standard per-token rates. For requests above 272,000 input tokens, OpenAI says input and cached-input rates double, while output costs 1.5 times the short-context rate for the whole request. Batch processing costs half the standard rate; Fast mode costs twice as much. Ultrafast is listed at three times standard pricing through Feb. 28, 2027, then six times from March 1, 2027. Eligible background requests have lower rates for two-hour and eight-hour completion windows.

OpenAI’s documentation also flags an API behavior change: from GPT-6.2, editing conversation history mid-session drops the model’s earlier reasoning, which may lower answer quality and cause it to use more tokens repeating prior work. Requests that set reasoning effort to “none” are rejected. The source report says rate limits vary by account tier, and that 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.

Limits of the Early Evidence

The available evidence is an early snapshot from Nov. 17, 2026. Artificial Analysis has not published an Intelligence Index score for Astra for Law, and the material available to the source report does not say what changes in that legal configuration. It also does not establish how the tested model settings translate to every use case or customer bill.

OpenAI’s launch post was not accessible to the source report, so it could not confirm which ChatGPT plans include Astra or Sol. The provided material also gives only a partial account of test-by-test results and does not include the full safety report’s evidence or an independent assessment of OpenAI’s cybersecurity and self-improvement classifications. These points should not be treated as settled by the headline scores.

What Developers Should Check

Developers considering either model can compare the published API rates with their expected prompt lengths, output volumes, and processing tier; long prompts and faster service can change the bill substantially. They will also need to test task quality, hallucination rates, and coding performance against their own workflows. For cybersecurity-related use, OpenAI’s stated access restrictions and vetting requirements are part of the current release conditions.

Further details may come through updates to OpenAI’s model and pricing pages, fuller release information, and additional independent evaluations. The next useful evidence will include a score for Astra for Law, clarification of its differences from Astra, and results showing how Astra and Sol perform across more real-world deployments. ChatGPT plan availability remains unconfirmed in the material used for this report.

Key Questions

How much do GPT-6.2 Astra and Sol cost?

OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens. Sol is $2 per million input tokens and $10 per million output tokens. Long requests, processing speed, and other pricing options can change the final cost.

Which model scored higher in the independent index?

Artificial Analysis scores Astra Max at 62 and Sol at 57. Claude Fable 5.2 leads the headline comparison at 65. Astra’s unrounded score of 62.14 is slightly below Fable 5.2 Xhigh at 62.22, so Astra’s model page ranks it third among 91 configurations.

What does OpenAI say about cybersecurity risk?

OpenAI’s system card rates both models Critical for cybersecurity, the higher of the two risk thresholds it uses. The company says more permissive cyber access requires extra vetting. This is OpenAI’s assessment, not an independent rating described in the source material.

Is Astra for Law a separate model?

OpenAI lists GPT-6.2 Astra for Law as a legal model configuration, with model ID gpt-6.2-astra-law. The available pages do not explain how it differs from Astra, and Artificial Analysis had not assigned it an Intelligence Index score as of Nov. 17, 2026.

Are Astra and Sol available in ChatGPT plans?

The source report could not confirm which ChatGPT plans include the new models, because OpenAI’s launch post was not accessible to it. The API documentation and pricing information do not by themselves settle plan availability.

Source: ThorstenMeyerAI.com

You May Also Like

Postgres rewritten in Rust, now passing 100% of the Postgres regression tests

The Postgres database engine has been rewritten in Rust and now successfully passes 100% of its regression tests, marking a significant milestone.

The Deployment Company, Back to the 70s, Apple and Intel

OpenAI launches The Deployment Company to embed AI in businesses, echoing 1970s computing, with backing from major firms; industry shifts focus to enterprise AI deployment.

They Don’t Break In, They Switch Your VPN Off: The Cisco Firewall Flaw Attackers Already Use, And How To Check Yours

Cisco says attackers are exploiting CVE-2026-76418, which can restart certain ASA and FTD firewalls through exposed SSL VPN services.

Meta to receive $3.3B in tax breaks for its $10B Louisiana data center

Meta’s $10B Louisiana data center, Hyperion, will get $3.3 billion in tax breaks, raising questions about public subsidies for tech infrastructure.