OpenAI launched GPT-6.2 Astra and Sol: the everyday work model drops 20% in price
Direct answer
OpenAI launched GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026, in the API, in ChatGPT Work and in Codex for Plus, Pro, Enterprise and Edu users. Astra costs $8 per million input tokens and $40 output, 20% less than GPT-6 Astra, and Sol holds at $2 input and $10 output. Astra matches Claude Fable 5.2's list price ($8/$40) and ranks second on the independent Artificial Analysis index, with 62 points against Fable 5.2's 65. Real savings are measured per finished task, and independent measurements put them between 36% and 50%, below OpenAI's headline.
What OpenAI launched on November 17
- Two models, one context: Astra is the model for demanding work and Sol the model for everyday work. Both have a 1.05-million-token context, a 128,000-token maximum output per response, and take text and images and return text. The API IDs are gpt-6.2-astra and gpt-6.2-sol.
- Astra drops 20% and cache drops 60%: GPT-6 Astra cost $10 input and $50 output; GPT-6.2 Astra costs $8 and $40. Cached input falls from $1.00 to $0.40 per million tokens. Sol keeps GPT-6.1 Sol's prices: $2 input, $0.10 cached and $10 output.
- The pricing fine print: the standard rates apply to requests up to 272,000 input tokens. Above that threshold, input and cache rates double and output costs 1.5 times as much for the whole request. Batch mode costs 50% less than standard.
- Available from day one: both models arrived the same day in the API, in ChatGPT Work and in Codex for the paid Plus, Pro, Enterprise and Edu plans. Free users are outside this launch list.
What the benchmarks say (and whose figures they are)
- Artificial Analysis, independent measurement: Astra scores 62 on the Intelligence Index, second behind Claude Fable 5.2 at 65. Sol scores 57. This is the launch's most useful independent reference, because it compares models with the same yardstick.
- OpenAI's figures are the vendor's: OpenAI claims Astra beats GPT-6 Astra by 5.2 points on Agents' Last Exam at under a third of the cost per task, and Claude Fable 5.2 by 5.5 points at 5% of the cost. These are OpenAI's claims, not independent measurements.
- Real savings, per independent measurements: Artificial Analysis puts Astra's cost per task 36% below GPT-6 Astra ($4.00 against $6.24), and Vals AI puts it about 50% below ($10.68 against $21.28). The savings are real and smaller than the headline.
- Sol performs close to Astra at a quarter of the price: OpenAI claims Sol lands within 0.5 points of Astra on Cognition's FrontierCode 2.0, at about a fifth of the cost, and ahead of GPT-6 Astra on its LifeVal test at about a seventh of the cost. These are OpenAI's figures; the lesson is to test Sol on your tasks before paying for Astra.
The safety signal that comes with the launch
- System card the same day: OpenAI published the GPT-6.2 system card alongside the models. Under its Preparedness Framework, both models sit at Critical for cybersecurity and High for biological and chemical risks.
- First at the High self-improvement threshold: OpenAI says Astra is the first released model to reach the High threshold on its own AI self-improvement scale, a level OpenAI itself classifies as a risk. This is the vendor's statement, and it reads as a signal that model capability is rising faster than controls.
- Frontier pricing converges: Astra arrives at the same list price as Claude Fable 5.2, launched eight days earlier. Competition between labs now plays out in price per task and independent benchmarks, not in announcements.
- What to do in your business this week: run the same task on GPT-6.2 Astra, GPT-6.2 Sol and your current model, and compare total cost per finished task, including cached tokens. If Sol solves your everyday tasks, the savings arrive without changing provider.
Note: this article summarizes OpenAI's announcement of November 17, 2026, OpenAI's API pricing page and coverage by The Decoder, Artificial Analysis, Vals AI, TechTimes and ArtificialWatch. Prices are OpenAI's standard rates for requests up to 272,000 input tokens. The independent benchmark figures are from Artificial Analysis and Vals AI; the performance and cost-per-task comparisons are OpenAI's claims. OpenAI did not announce GPT-6 Luna in this launch.
Want AI that works for your business?
Message us on WhatsApp and let's talk about automation and AI agents for your business.
WhatsApp inicIARelated articles
- OpenAI launched GPT-6.1 Terra at half price: $1 per million tokens and a ranking Anthropic still leads
- Claude Fable 5.2 and Mythos 5.2: 20% cheaper and up to 30% faster
- GPT-6 arrived with a smart interface and ads in chat: your customers' new storefront
Frequently asked questions
How much does GPT-6.2 Astra cost?
$8 per million input tokens, $0.40 per million cached input tokens and $40 per million output tokens, at standard rates for requests up to 272,000 tokens. That is 20% less than GPT-6 Astra on input and output, and 60% less on cache.
How much does GPT-6.2 Sol cost?
$2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens. It keeps GPT-6.1 Sol's prices and costs a quarter of Astra per token.
Where are they available?
In the OpenAI API (gpt-6.2-astra and gpt-6.2-sol), in ChatGPT Work and in Codex, for Plus, Pro, Enterprise and Edu users, from November 17, 2026. Free users are outside the launch.
Which performs better, Astra or Claude Fable 5.2?
On the independent Artificial Analysis index, Claude Fable 5.2 scores 65 points and GPT-6.2 Astra 62, with Sol at 57. Both models share the same list price: $8 input and $40 output per million tokens.
