Moving from GPT-6 Astra and GPT-6.1 Sol to GPT-6.2
· openai, models, api
Update your model IDs and API key allowlist. See the new Astra prices, Sol prices, long-context threshold and a Responses request.
November 18, 2026 update: later the same day TopxAI moved Claude, OpenAI and Grok models to one route per provider at 80% of the provider's list price. gpt-6.2-astra now costs $6.40 input / $32 output and gpt-6.2-sol $1.60 / $8 per million tokens (Astra, Sol). The table below shows the prices at publication.
On the OpenAI route, use gpt-6.2-astra wherever your application selected gpt-6-astra, and gpt-6.2-sol wherever it selected gpt-6.1-sol. Auto-route keys can use either model. If your key has a model allowlist, add the corresponding new ID in API Keys. A request with a retired ID returns HTTP 403 and the error code permission_denied.
Send the first request
Keep your base URL and credentials. In Codex, update the model entry in ~/.codex/config.toml. To try Sol directly, send the following request. Use gpt-6.2-astra in the same command for Astra.
curl https://api.topxai.xyz/v1/responses \
-H "Authorization: Bearer $TOPXAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"gpt-6.2-sol","input":"Name two ways to check a database migration.","reasoning":{"effort":"medium"}}'
Both models accept text and image inputs, with a 1,050,000-token context window, up to 922,000 input tokens and 128,000 output tokens. Use Responses for function calling. When migrating a saved client configuration, set reasoning effort to low or higher. Remove temperature, top_p, logprobs and top_logprobs settings and any message.output_text.logprobs entry in Responses include. OpenAI migration guide
What a request costs
Default short-context rates in USD per million tokens follow. Compared with the previous Astra, input, output and default cache writes each cost 20% less, and cache reads cost 60% less. Sol's default rates stay the same.
| Model | Route | Input | Cache read | Cache write | Output |
|---|---|---|---|---|---|
| Astra | Shared | $4 | $0.20 | $5 | $20 |
| Astra | Official | $7.20 | $0.36 | $9 | $36 |
| Sol | Shared | $1 | $0.05 | $1.25 | $5 |
| Sol | Official | $1.80 | $0.09 | $2.25 | $9 |
At exactly 272,000 input tokens, including cached tokens, the table still applies. From 272,001 input tokens, the full request uses double the input and cache rates and 1.5 times the output rate. For example, at publication Astra's shared route then cost $8 per million ordinary input tokens and $30 per million output tokens; at today's price it is $12.80 and $48. Charges for earlier requests keep their recorded prices. OpenAI Astra pricing · OpenAI Sol pricing
Models: gpt-6.2-astra, gpt-6.2-sol · Models & pricing
Related
- GPT-6.2 Astra API pricing
- GPT-6.2 Sol API pricing
- Codex CLI custom model provider: model_providers in config.toml for TopxAI
- GPT-6.1 Terra on TopxAI
- Moving to GPT-6.1 Sol on TopxAI
- Abliterated models on TopxAI: the November 10 prices and how to call them
Other languages: 简体中文