OpenAI launches GPT-6.2 Astra and Sol with lower Astra API prices
Edited and reviewed by AI Ping Editorial Team
3-line summary
- Both models are available in ChatGPT Work, Codex and the API. The regular Chat tab is outside this rollout.
- Astra’s standard API rates are $8 for input and $40 for output per million tokens.
- Sol keeps GPT-6.1 Sol’s standard API rates.
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Nov. 17. ChatGPT Plus, Pro, Enterprise and Edu users can select the new models in Work and Codex. OpenAI’s help center says the rollout leaves the regular Chat tab out. Both models are also available through the OpenAI API.
For API customers, Astra’s standard input and output rates are 20% lower than those for GPT-6 Astra. Sol keeps the rates of GPT-6.1 Sol. Workspace administrators control access on Enterprise and Edu. Pro 500 subscribers get the faster Ultrafast mode for both models. On Pro 200 it starts with GPT-6.2 Sol. OpenAI says eligible Enterprise and Edu users can get the same access once their administrator enables it.
Work documents and Codex tasks
OpenAI positions Astra for complex assignments and Sol for everyday work. One task the company highlighted is building slides that follow the layout and formatting of an existing deck or a screenshot. In the API, both models accept text and image inputs and generate text. Work can turn a model’s output into a document through its tools, while the model endpoints specify what each request can send and receive.
In OpenAI’s GDPval v2 evaluation, the new Astra scored 6.2 percentage points higher than GPT-6 Astra while using 22% fewer output tokens. The evaluation uses multi-step professional assignments based on actual projects, with the associated files and messages supplied for the task.
Another internal evaluation deliberately gave the models coding jobs with obstacles, such as a missing file. GPT-6.2 Astra omitted a known obstacle from its answer in 22.28% of trials, up from 11.09% for GPT-6 Astra. OpenAI says the instructions used in Codex reduced such omissions in tests. Those instructions are specific to Codex and are absent from direct API calls.
Astra’s lower rates and Sol’s unchanged prices
Astra’s standard API prices per million tokens are $8 for ordinary input, $0.40 for cached input and $40 for output. The preceding Astra rates were $10, $1 and $50, respectively. Cached input thus gets the steepest cut at 60%. Sol costs $2, $0.10 and $10, respectively, the same rates as GPT-6.1 Sol.
Take a hypothetical set of standard-rate requests totaling a million uncached input tokens and 200,000 output tokens. At those counts, the token bill falls from $20 on the previous Astra to $16 on the new one. The same counts come to $4 on the new Sol. The calculation excludes caching and separate tool fees. Actual costs depend on how much input gets reused and how much output a model produces for the work.
The cached input price applies when a request reuses the opening portion of a prompt that OpenAI already stored. A reusable prefix has to reach at least 1,024 tokens on these models. OpenAI counts tokens written to the cache separately from ordinary input and bills them at the write rate. Under standard short-context pricing, a 30-minute write costs $10 per million tokens on Astra and $2.50 on Sol. For 12-hour storage in the Responses API, those write rates rise to $20 and $5.
Higher rates for longer requests and faster processing
For either model, a request with more than 272,000 input tokens is billed at twice the usual input and cache rates and 1.5 times the output rate. Those rates apply to the full request. Both models have a 1.05 million-token context window, a separate limit on how much the model can handle at once. OpenAI lists a maximum of 922,000 input tokens and 128,000 output tokens for each model.
The API’s Ultrafast tier costs three times standard rates for both models through Feb. 28, 2027, Pacific time. The multiplier rises to six on March 1.
Sources
- Introducing GPT-6.2 Astra and Sol — OpenAI
- GPT-6 and other models in ChatGPT — OpenAI
- Model Release Notes — OpenAI
- GPT-6.2 Astra Model — OpenAI
- GPT-6.2 Sol Model — OpenAI
- GPT-6 Astra Model — OpenAI
- GPT-6.1 Sol Model — OpenAI
- Prompt caching — OpenAI
- GPT-6.2 System Card — OpenAI