Model pricing
gpt-6.2-sol
GPT lane (OpenAI protocol)
Models on the GPT lane: same account and key as the Claude lane (OpenAI protocol at /v1). astra is the high-performance tier, sol the balanced daily driver — a natural fit for Codex and everyday engineering.
Unit: USD per million tokens
| Billing item | Base rate | Your rate (standard group) |
|---|---|---|
| Input | $2.00 | $0.13 |
| Output | $10.00 | $0.65 |
Output bills at 5× the input rate; cached prompt reads bill at a lower rate. Rates on this page come live from the pricing ledger; your authoritative bill is Usage in the workspace.
Official base rates: OpenAI
“Tokens are billed at the chosen model's input and output rates.”
How to connect
One key, both protocols: Claude models take the ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN environment variables; GPT models just need the base_url in the OpenAI SDK.