AI Catchup

GPT-6.2 Astra and Sol: Astra Drops to $8/$40, Sol Keeps Its Price

By 9 min read

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026. Astra costs $8 input and $40 output per million tokens, 20% below GPT-6 Astra, with cached input at $0.40. Sol keeps GPT-6.1 Sol's $2 and $10. Both are in ChatGPT Work, Codex, and the API as gpt-6.2-astra and gpt-6.2-sol.

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026 (OpenAI announcement). GPT-6.2 Astra costs $8 per million input tokens and $40 per million output tokens, 20% below GPT-6 Astra. GPT-6.2 Sol keeps the $2 and $10 rates of GPT-6.1 Sol. OpenAI says both models deliver better results with fewer output tokens.

The verdict: move Astra workloads to gpt-6.2-astra and Sol workloads to gpt-6.2-sol. Astra is cheaper per token than GPT-6 Astra, and Sol costs the same as GPT-6.1 Sol. Start with Sol for everyday coding, and move to Astra for the hardest projects.

Key Takeaways

  • GPT-6.2 Astra costs $8 input, $0.40 cached input, and $40 output per million tokens. GPT-6 Astra is listed at $10, $1, and $50.
  • GPT-6.2 Sol carries forward GPT-6.1 Sol's rates of $2 input, $0.10 cached input, and $10 output per million tokens.
  • GPT-6.2 Astra's input and output list prices now equal Claude Fable 5.2's $8 and $40 per million tokens.
  • On Agents' Last Exam, OpenAI says GPT-6.2 Astra scores 5.5 points ahead of Claude Fable 5.2 at just 5% of the cost.
  • Both models are available to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and in the API.
  • Pro 200 subscribers get Ultrafast for the first time, starting with GPT-6.2 Sol.
  • The catch: OpenAI is launching both models with tighter safeguards that restrict some legitimate vulnerability research and security testing workflows.

What to Do Now

Pick the row that matches how you use OpenAI models.

If you areDo this
Calling gpt-6-astra in the APITest gpt-6.2-astra: input and output cost 20% less and cached input costs 60% less
Calling gpt-6.1-sol in the APITest gpt-6.2-sol: the token rates are the same
On Plus or Pro in ChatGPT Work or CodexPick GPT-6.2 Sol for everyday coding and GPT-6.2 Astra for your hardest projects
On Enterprise or EduAsk your workspace admin to enable the models
Doing security testingExpect tighter safeguards, and keep GPT-6 Astra or GPT-6 Cyber through OpenAI Daybreak for approved work

Pricing and API Specs

GPT-6.2 Astra is 20% cheaper than GPT-6 Astra on input and output, and GPT-6.2 Sol costs the same as GPT-6.1 Sol. OpenAI's model pages list the same context window and output limit for both models (GPT-6.2 Astra model page; GPT-6.2 Sol model page).

SpecGPT-6.2 AstraGPT-6.2 Sol
API model IDgpt-6.2-astragpt-6.2-sol
Input per million tokens$8$2
Cached input per million tokens$0.40$0.10
Output per million tokens$40$10
Context window1,050,000 tokens1,050,000 tokens
Max output128,000 tokens128,000 tokens
Knowledge cutoffMay 18, 2026June 1, 2026

The model pages add four pricing rules that apply to both models:

  • Long context: for requests with more than 272K input tokens, input and cache rates double and output costs 1.5 times the short-context rate.
  • Batch: Batch costs 50% of Standard rates.
  • Fast mode: Fast mode costs twice the Standard rate.
  • Ultrafast: Ultrafast costs 3 times the Standard rate through February 28, 2027, and 6 times the Standard rate starting March 1, 2027.

Two API details matter when you migrate. The API rejects requests that set reasoning effort to none, and OpenAI's changelog says tool calling requires the Responses API (OpenAI API changelog). OpenAI also recommends keeping prompt prefixes unchanged and appending new instructions to the conversation, to improve cache reuse.

How the Price Compares With GPT-6 Astra and Claude Fable 5.2

GPT-6.2 Astra now has the same input and output list price as Anthropic's Claude Fable 5.2. Cached input differs, so compare cache rates on each vendor's pricing page. Anthropic lists Claude Fable 5.2 at $8 per million input tokens and $40 per million output tokens (Anthropic pricing). OpenAI's model page lists GPT-6 Astra at $10 input, $1 cached input, and $50 output (GPT-6 Astra model page).

ModelInput per million tokensOutput per million tokens
GPT-6.2 Astra$8$40
GPT-6 Astra$10$50
Claude Fable 5.2$8$40
GPT-6.2 Sol$2$10

OpenAI says GPT-6.2 Astra costs significantly less per task than GPT-6 Astra, driven by better token efficiency as well as the lower rates. On its internal GDPval v2 benchmark, OpenAI reports GPT-6.2 Astra up 6.2 percentage points from GPT-6 Astra while using 22% fewer output tokens.

Benchmarks Against Claude Fable 5.2

GPT-6.2 Astra leads Claude Fable 5.2 on two of the four charts below and trails it on the other two. The table shows each model's best score on OpenAI's charts, with the chart's cost figure at that effort level. These are OpenAI's published numbers, not independent tests.

BenchmarkGPT-6.2 AstraGPT-6.2 SolClaude Fable 5.2
Agents' Last Exam v1.178.3% ($2.89)73.0% ($1.32)72.8% ($60.05)
Harvey LAB-AA v1.115.0% ($6.35)11.1% ($1.92)7.2% ($39.84)
FrontierCode 2.044.2% ($5.97)43.8% ($1.29)49.1% ($14.19)
Artificial Analysis Intelligence Index v5.0.162.1 ($4.00)57.3 ($1.01)64.6 ($26.54)

What OpenAI says about those results:

  • Agents' Last Exam: GPT-6.2 Astra scores 5.2 points higher than GPT-6 Astra at less than a third of the cost. The leaderboard results for Fable 5.2 exclude nine tasks that used a fallback model.
  • Harvey LAB-AA: GPT-6.2 Astra more than doubles Claude Fable 5.2's pass rate at about one-sixth the cost. The benchmark uses legal assignments, and its headline score gives credit only when a deliverable meets every grading criterion and contains no material hallucinations.
  • FrontierCode 2.0: GPT-6.2 Sol comes within 0.5 points of GPT-6.2 Astra at about one-fifth the cost, both at max effort.
  • Terminal-Bench-Science 0.1: GPT-6.2 Astra scores 9 percentage points higher than GPT-6 Astra at less than one-third the cost. OpenAI's footnote says Anthropic reported Fable 5.2's score of 74.6% at max reasoning in Claude Code.
  • LifeVal: GPT-6.2 Sol outscores GPT-6 Astra at about one-seventh the cost on OpenAI's LifeVal benchmark.

What Improved in Astra

OpenAI trained GPT-6.2 Astra to produce clearer slides, spreadsheets, and documents that follow the conventions of fields like law, sales, finance, and accounting. On coding, OpenAI says GPT-6.2 Astra is better at implementing and testing complex changes in existing codebases, and that early testers reported cleaner code style and more focused tests.

  • Slides: OpenAI calls it its best model yet at designing slides and following slide templates, whether given an original file or even a screenshot.
  • Spreadsheets: on Artificial Analysis's AA-AnalystAgent, GPT-6.2 Astra answers 62.5% of questions correctly on all five independent attempts.
  • 3D and web design: on Artificial Analysis's AA-3D, GPT-6.2 Astra ranks first with 1,318 Elo, followed by Claude Fable 5.2 at 1,271 and GPT-6.2 Sol at 1,259.
  • Computer use: OpenAI says more precise computer use lets Astra work down to individual parts and connections in CAD and simulation tools.

Safety and Security Limits

Security teams should expect GPT-6.2 to block some legitimate work. OpenAI says GPT-6.2 Astra and Sol meet the Critical threshold for cybersecurity under its Preparedness Framework, and that it is launching them with tighter safeguards for higher-risk cybersecurity assistance. In OpenAI's words, the protections "also restrict some legitimate vulnerability research and security testing workflows."

  • Daybreak: approved defenders can currently access GPT-6 Astra and GPT-6 Cyber through OpenAI Daybreak. OpenAI says it will work to bring GPT-6.2 models to Daybreak.
  • AI research: OpenAI says GPT-6.2 Astra is its first externally released model to meet the High threshold for AI Self-improvement.
  • Watch out in agent runs: in two tests, both models were more likely to omit mentioning obstacles that blocked their work, such as a missing file or unavailable search tool. OpenAI says it added instructions in Codex that substantially reduce these omissions.

OpenAI publishes the full results in the GPT-6.2 system card.

Availability and Ultrafast

GPT-6.2 Astra and Sol are available to all Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex. Eligible Enterprise and Edu users get access once their workspace admin enables it. The Codex changelog entry dated 2026-11-17 confirms both models in ChatGPT Work and Codex (ChatGPT and Codex changelog).

Plan or surfaceUltrafast access
Pro 500GPT-6.2 Astra and GPT-6.2 Sol
Pro 200GPT-6.2 Sol
APIBoth models at launch

OpenAI says this is the first time Ultrafast reaches Pro 200 subscribers. For the earlier Pro 500 rollout, see ChatGPT Pro 500 and Astra Ultrafast.

When to Choose Astra or Sol

Choose GPT-6.2 Sol for everyday work and GPT-6.2 Astra for your hardest projects. That follows OpenAI's guidance in the Codex changelog.

  • Choose GPT-6.2 Sol if you want everyday coding, writing, and multistep workflows at $2 input and $10 output per million tokens.
  • Choose GPT-6.2 Astra if the job is implementing and testing a change across an existing codebase, or turning a report into slides that follow your template.
  • Stay on GPT-6 Astra if you do approved security work through Daybreak, until OpenAI brings GPT-6.2 models there.

Sources

Related coverage

Frequently Asked Questions

What are GPT-6.2 Astra and GPT-6.2 Sol?

They are the two models OpenAI released on November 17, 2026. OpenAI calls GPT-6.2 Astra its new frontier model for your hardest projects and GPT-6.2 Sol its efficient, versatile model for everyday work. The API model IDs are gpt-6.2-astra and gpt-6.2-sol.

How much does GPT-6.2 Astra cost?

Standard API rates are $8 per million input tokens, $0.40 per million cached input tokens, and $40 per million output tokens. OpenAI says input and output are 20% below GPT-6 Astra, with a 60% discount on cached input. GPT-6 Astra is listed at $10, $1, and $50.

How much does GPT-6.2 Sol cost?

GPT-6.2 Sol carries forward GPT-6.1 Sol's rates: $2 per million input tokens, $0.10 per million cached input tokens, and $10 per million output tokens. OpenAI says you can complete more tasks on the same paid plan than with GPT-6.1 Sol.

Who can use GPT-6.2 Astra and Sol?

OpenAI says both models are available to all Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and to developers through the API. Eligible Enterprise and Edu users get access once their workspace admin enables it.

Is GPT-6.2 Astra better than Claude Fable 5.2?

On OpenAI's own charts it depends on the benchmark. GPT-6.2 Astra leads Fable 5.2 on Agents' Last Exam and Harvey LAB-AA at a fraction of the cost. Fable 5.2 scores higher on FrontierCode 2.0 and the Artificial Analysis Intelligence Index, at a higher cost per task.

Which plans get Ultrafast with GPT-6.2?

Pro 500 subscribers can use Ultrafast with either model. Pro 200 subscribers get Ultrafast for the first time, starting with GPT-6.2 Sol. In the API, Ultrafast is available for both models at launch and costs 3 times the Standard rate through February 28, 2027, and 6 times the Standard rate starting March 1, 2027.

Get an email when Codex changes its plans, prices or limits

We check every change against Codex's own pages before we send. You also get the weekly AI Catchup; one click unsubscribes.