Urdu Pro
Pakistan News, Tech, Sports & Showbiz

OpenAI Launches GPT-6.2 Astra and Sol: Cheaper, Faster Models Land as Anthropic IPO Prices Today

OpenAI has launched GPT-6.2 Astra and Sol, two new artificial intelligence models that promise better results at lower prices — and the cheaper of the pair is already being compared with rivals that cost far more to run.

The company announced GPT-6.2 Astra and Sol on Tuesday 17 November, and both models are available now in ChatGPT Work and Codex for paying users, as well as through OpenAI’s API. Astra is the flagship model for the hardest professional work, while Sol is the efficient everyday model for coding, writing and agents. The launch comes just as rival Anthropic prepares to price its blockbuster stock market debut — a week in which the AI race is being fought on price as much as on performance.

GPT-6.2 Astra and Sol: What OpenAI Launched

GPT-6.2 Astra and Sol arrive as a pair, and OpenAI is pitching them as a clear step up from the GPT-6 generation. Astra is described by the company as its frontier model “for your hardest projects”, built for demanding work across code, research and professional documents. Sol is the “efficient, versatile model for everyday work” — the one most people will actually use day to day.

Both models are rolling out to Plus, Pro, Enterprise and Edu subscribers in ChatGPT Work and Codex, and to developers through the API under the model names gpt-6.2-astra and gpt-6.2-sol. OpenAI says both models finish demanding tasks using fewer output tokens than their predecessors, so users should get more done within the same usage limits. The models carry a context window of 1.05 million tokens and a maximum output of 128,000 tokens. Sol’s knowledge reaches to 1 June 2026, while Astra’s cutoff is 18 May 2026.

UrduPro reported earlier this month that GPT-6.2 Astra was expected to launch, and the release confirms that reporting — with confirmed prices and benchmark scores that the earlier preview could only speculate about.

GPT-6.2 Astra and Sol Pricing: Astra 20% Cheaper, Sol Unchanged

The headline for developers is the price. GPT-6.2 Astra costs $8 per million input tokens and $40 per million output tokens, with cached input at $0.40 — 20% below the $10 and $50 rates of GPT-6 Astra. Sol keeps exactly the rates of GPT-6.1 Sol: $2 per million input tokens, $0.10 for cached input and $10 for output.

That makes Sol the value story of the launch. It costs the same as before, yet OpenAI says it comes within half a point of Astra on Cognition’s FrontierCode 2.0 benchmark — a test of whether changes to real codebases are good enough to merge — at about one-fifth of the cost. On LifeVal, a test of challenging everyday tasks, Sol even outscores GPT-6 Astra at about one-seventh of the cost.

OpenAI is also widening access to Ultrafast, its highest-speed mode. Pro 500 subscribers can use Ultrafast with either model, and Pro 200 subscribers get Ultrafast for the first time, starting with Sol. The company is cutting Ultrafast pricing by 50% for the next three months.

How GPT-6.2 Astra and Sol Score on Benchmarks

OpenAI’s own benchmark figures show Astra improving across the board. On Agents’ Last Exam, which tests multi-step professional work, Astra scores 5.2 points above GPT-6 Astra at less than one-third of the cost per task. On Terminal-Bench-Science 0.1, a science research test, it improves by nine percentage points. On GDPval v2, built from consulting, finance and design assignments, Astra improves 6.2 points while using 22% fewer output tokens.

Independent testing tells a similar story, with a caveat. Artificial Analysis, which runs its own benchmark suite on new models, puts Astra at 62 on its Intelligence Index, up from 57 for GPT-6 Astra, and Sol at 57, up from 55 for GPT-6.1 Sol. Anthropic’s Claude Fable 5.2 still leads the index at 65, with Claude Opus 5.5 at 59. But the cost gap is striking: running the index costs $4.00 with Astra and just $1.01 with Sol, against $10.04 for Opus 5.5 and $26.54 for Fable 5.2. Sol scores two points behind Opus 5.5 for roughly a tenth of the cost.

Tighter Safety Rules for GPT-6.2 Astra and Sol

The launch also comes with stricter safety limits. OpenAI rates both models at the Critical level for cybersecurity under its Preparedness Framework, and Astra is the first externally released OpenAI model rated High for AI self-improvement — the ability of a model to improve itself. The company says both models are launching “with tighter safeguards for higher-risk cybersecurity assistance”.

The caution follows OpenAI’s own testing: on its ExploitBench test, Astra jumps 22 points over GPT-6 Astra, and Sol lands close to Astra and well above GPT-6.1 Sol. Security researchers doing legitimate vulnerability work may find the new models refuse more often than the old ones, though OpenAI says it is working to bring the models to Daybreak, its programme that gives approved security teams access with fewer refusals.

Early Reaction to GPT-6.2 Astra and Sol — and the Anthropic Timing

The first hours of reaction were split. On Reddit, many users praised the pricing, with one commenter writing that Sol matches Opus 5.5 “at a fraction of the price”, and others saying the release could make them switch from Claude. But some early testers were critical: AI investor Matt Shumer said after first-hour testing that the models were “essentially unusable” and that he would not recommend them — a single early impression, not a verdict. Others complained that OpenAI did not reset usage limits for launch day, and that the rollout was uneven, with some users seeing only one of the two models at first.

The timing is no accident. Anthropic, the company behind Claude, is in the middle of its IPO roadshow and is expected to set its final share price today, 18 November, at $70 to $78 a share — a deal that could value the company at up to $1.4 trillion, with trading under the ticker ANT starting on Thursday. OpenAI’s launch, with its aggressive pricing, is a reminder that the contest between the two AI labs is now being fought in public, on price, on benchmarks — and on the stock market.

For users, the question now is whether the cheaper models feel better in daily work. The API prices are easy to check and hard to argue with. The subscription experience will take a few days of normal use to judge.

Related on UrduPro: Nvidia reported record quarterly revenue of $115.8 billion as its data-centre business boomed.

GPT-6.2 Astra and Sol
Written by UrduPro Editorial Team
Articles on UrduPro are researched and reviewed by our editorial team before publication. See our Editorial Policy and Contact pages for how we work and how to send corrections.
UrduPro Editorial Team
UrduPro Correspondent

Articles on UrduPro are researched and reviewed by our editorial team before publication. See our Editorial Policy and Contact pages for how we work and how to send corrections.

Leave a Reply

Your email address will not be published. Required fields are marked *

URDU PRO

Pakistan News, Tech, Sports & Showbiz

Get the top stories daily

Breaking news and updates, straight to your inbox.

© 2026 Urdu Pro. All rights reserved.