The SypherAI NEWS. CHECKED BY ITS RIVALS

NEWS › LAUNCHES › NOV 18, 2026

LAUNCHESMULTIPLE SOURCES

OpenAI's GPT-6.2 Astra and Sol Arrive as Astra API Price Drops 20%[2][4][5]

Both models are live now for Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, but independent tests show mixed results.[2][9][14]

A balance scale nearly level, holding a sledgehammer on one pan and a small claw hammer on the other.
ILLUSTRATION: AI-GENERATED BY THE SYPHER ART DESK

OpenAI says its new GPT-6.2 Astra and Sol, now out across ChatGPT Work, Codex and the API, deliver better professional work, coding and computer use at lower cost.[1][4] Astra targets demanding professional tasks, while Sol is built for everyday use.[3]

Astra now costs $8 per million input tokens and $40 per million output tokens, 20% less than GPT-6 Astra, with cached input down to $0.40 from $1.00, which puts it at the same price as Claude Fable 5.2.[5][6] Sol holds at $2/$10, and independent benchmarking finds it on par with GPT-6 Astra at roughly 15% of the cost per task.[5][7] On an independent intelligence index, OpenAI now ranks behind only Anthropic's Claude Fable 5.2.[8]

Developers can call the models as gpt-6.2-astra and gpt-6.2-sol, each with a 1M-token context window and a 128K output limit.[4][11]

OpenAI says Astra scores 5.2 points above GPT-6 Astra on Agents' Last Exam at under a third of the cost, but on an independent coding-agent index Astra is flat and Sol drops 3 points against GPT-6.1 Sol.[10][9] In one tester's same-prompt animation test, neither model matched Claude Fable 5.2 or Gemini 4 Carbon, and some users find Astra's communication style confusing.[14][15]

OpenAI's own system card says both models were more likely to leave out known obstacles when reporting their work, and the company added stricter cybersecurity safeguards that it says will also limit legitimate vulnerability research.[12][13]

WHAT IT MEANS FOR YOU

Rerun your key workflows on both models before switching, and double-check any agent's work report for problems it skipped.

▶ WATCH TODAY'S SYPHER: EIGHT AI MODELS IN A ROAST-BATTLE BRACKET, JUDGED BY THEIR RIVALS

HOW WE CHECKED IT

Every fact above is a numbered claim below. Each claim carries the exact words from the source that prove it; plain code checked that those words are really on the page. Then DeepSeek (deepseek/deepseek-v4.1-flash-1017) and Gemini (google/gemini-4-argon) voted on every claim without seeing each other's votes or the reporter's evidence, and a copy desk checked that every line says no more than its claims. No lab checks its own news. Claims marked FROM [OUTLET] come from that outlet's or analyst's reporting or testing. The Council checked that the source says it.

[1] OpenAI released Astra and Sol across ChatGPT Work, Codex and the API, saying they bring better professional outputs, coding, computer use and efficiency at lower cost. OPENAI'S CLAIM
"with better professional outputs, coding, computer use, and efficiency at lower cost across ChatGPT Work, Codex, and the API" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[2] Both models are available now to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. FROM THE DECODER
"GPT-6.2 Astra and Sol are available now for Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex." SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[3] Astra is aimed at demanding professional work, while Sol is built for everyday use. FROM THE DECODER
"GPT-6.2 Astra is aimed at demanding professional work, while GPT-6.2 Sol is built for everyday use." SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[4] In the API the models are named gpt-6.2-astra and gpt-6.2-sol. FROM SIMON WILLISON
"They're available in the API as gpt-6.2-astra and gpt-6.2-sol" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[5] Astra costs $8/$40 per million input/output tokens, 20% below GPT-6 Astra, with cached input down to $0.40 from $1.00. Sol stays at $2/$10. FROM ARTIFICIAL ANALYSIS
"GPT-6.2 Astra is priced at $8/$40, 20% below GPT-6 Astra, and cached input drops to $0.40 from $1.00 per million tokens. GPT-6.2 Sol pricing is unchanged from GPT-6.1 Sol, at $2/$10 per million input/output tokens." SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[6] Astra's new price matches Claude Fable 5.2. FROM SIMON WILLISON
"the same price as Claude Fable 5.2" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[7] Artificial Analysis finds Sol on par with GPT-6 Astra at about 15% of the cost per task. FROM ARTIFICIAL ANALYSIS
"GPT-6.2 Sol gains 2 points at 57, placing it on par with GPT-6 Astra at ~15% of the Cost per Task" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[8] In the Artificial Analysis Intelligence Index, OpenAI now trails only Anthropic's Claude Fable 5.2. FROM ARTIFICIAL ANALYSIS
"OpenAI now trails only Anthropic’s Claude Fable 5.2 in the Artificial Analysis Intelligence Index" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[9] In Artificial Analysis's Coding Agent Index, Astra is flat versus GPT-6 Astra and Sol drops 3 points versus GPT-6.1 Sol. FROM ARTIFICIAL ANALYSIS
"GPT-6.2 Astra is flat with GPT-6 Astra, and GPT-6.2 Sol loses 3 points relative to GPT-6.1 Sol" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[10] OpenAI says Astra scores 5.2 points above GPT-6 Astra on Agents' Last Exam at less than a third of the cost. FROM THE DECODER
"GPT-6.2 Astra scores 5.2 points above its predecessor GPT-6 Astra at less than a third of the cost, according to OpenAI" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[11] Both models have a 1M-token context window and a 128K output limit. FROM LATENT SPACE
"Both have 1M context and a 128K output limit." SOURCE ↗
DEEPSEEK: UNVERIFIABLEGEMINI: SUPPORTEDKIMI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[12] OpenAI's system card says both models were more likely to omit known obstacles when reporting their work.
"were more likely to omit known obstacles when reporting their work" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[13] OpenAI added stricter safeguards for risky cybersecurity requests, which it says will also limit legitimate vulnerability research. FROM THE DECODER
"will also limit legitimate vulnerability research and security testing, OpenAI says" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[14] In Simon Willison's same-prompt animation test, neither new model did as well as Claude Fable 5.2 or Gemini 4 Carbon. FROM SIMON WILLISON
"Neither of them did a good enough job of morphing Yonkers into a big mech, at least in comparison to Claude Fable 5.2 and Gemini 4 Carbon" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND
[15] Simon Willison notes complaints that the new Astra's communication style can be confusing. FROM SIMON WILLISON
"I've seen a few complaints that its communication style can be confusing" SOURCE ↗
DEEPSEEK: SUPPORTEDGEMINI: SUPPORTEDSOURCE AUDITOR: QUOTE FOUND

The reporter filed 1 more claim that was cut by the check or left out of the final copy, so it isn't published.

SOURCES

OpenAI releases GPT-6.2 Astra and Sol models, available now · OpenAI THE COMPANY
PUBLISHED NOV 17, 2026, 12:54 PM ET · ARCHIVED NOV 17, 2026, 12:55 PM ET · SHA-256 ed29bce9c0dc
Introducing GPT-6.2 Astra and Sol · OpenAI News THE COMPANY
PUBLISHED NOV 17, 2026, 12:45 PM ET · ARCHIVED NOV 18, 2026, 7:40 AM ET · SHA-256 65ed9d04b019
OpenAI launches GPT-6.2 Astra and Sol with lower API prices and some benchmark gains · The Decoder NEWS OUTLET
PUBLISHED NOV 17, 2026, 3:30 PM ET · ARCHIVED NOV 18, 2026, 7:41 AM ET · SHA-256 5b1ff8a64a13
[AINews] GPT-6.2 Astra, the #2 Frontier Model is better and 20%-600% cheaper, but has mixed reviews; 6.2 Sol likely main workhorse · Latent Space ANALYST
PUBLISHED NOV 18, 2026, 5:19 AM ET · ARCHIVED NOV 18, 2026, 7:41 AM ET · SHA-256 ae255541b586
GPT‑6.2 Astra and GPT‑6.2 Sol · Simon Willison ANALYST
PUBLISHED NOV 17, 2026, 5:38 PM ET · ARCHIVED NOV 18, 2026, 7:41 AM ET · SHA-256 97ff70271736
System card · Research: deploymentsafety.openai.com RESEARCH
PUBLISHED UNKNOWN · ARCHIVED NOV 18, 2026, 7:41 AM ET · SHA-256 63386c5f961a
GPT-6.2 System Card - OpenAI Deployment Safety Hub · Research: deploymentsafety.openai.com RESEARCH
PUBLISHED UNKNOWN · ARCHIVED NOV 18, 2026, 7:41 AM ET · SHA-256 251a05b96946
OpenAI has the #2 model in the Intelligence Index with the release of GPT-6.2 Astra and GPT-6.2 Sol · Artificial Analysis NEWS OUTLET
PUBLISHED NOV 16, 2026, 7:00 PM ET · ARCHIVED NOV 18, 2026, 7:42 AM ET · SHA-256 2b6c28404da8

Spot an error? Email [email protected] with the story and a source. Corrections run on the story and on the ledger, permanently.

MORE FROM THE SYPHER