GPT-6 API without a subscription: pick the right model, pay per token.

Live GPT-6 prices

Read straight from the catalog. A model that is temporarily unavailable stays listed with its price but cannot be called until its checks pass again.

Which GPT-6 model to use

• GPT-6 Sol is the everyday coding and chat model: the best balance of quality and price, and the default for Codex CLI and Cursor.
• GPT-6.1 Sol is the newer Sol revision at a similar price; try it first when you start a new project.
• GPT-6 Astra is the top tier for hard problems, long refactors and agents where one wrong step is expensive. It costs the most per token.
• GPT-6 Luna is the small, inexpensive model for routine edits, summaries and tests of your integration. Its price is close to list price.

Rule of thumb: start with Luna to test, run daily work on Sol, and move a task to Astra only after it fails twice. Compare the exact charge of each request in your dashboard.

Call GPT-6 in 30 seconds

Create a key, add $5 or more, and point any OpenAI client at https://api.model.sale/v1.

curl https://api.model.sale/v1/chat/completions \\
  -H "Authorization: Bearer $MODEL_SALE_KEY" \\
  -H "Content-Type: application/json" \\
  -d '{"model":"gpt-6-sol","messages":[{"role":"user","content":"Hello"}]}'

from openai import OpenAI
client = OpenAI(base_url="https://api.model.sale/v1", api_key="YOUR_KEY")
r = client.responses.create(model="gpt-6-sol", input="Hello")
print(r.output_text)

Common questions

What does GPT-6 cost per request? It depends on tokens, not on a fixed fee. A medium coding session of about 1M charged tokens costs roughly the price per 1M shown in the table above; see the Codex session cost guide.

Does it work with Codex CLI and Cursor? Yes. Use the base URL above with the Responses or Chat Completions endpoint; see the setup guides.

Is the requested model ever replaced? No. If a model is unavailable the request fails with a clear error instead of silently switching.