GPT-6.2 Astra
OpenAI's most capable model for demanding reasoning and coding, building and revising software, operating computers, and working with documents on long projects, with tool calling and structured outputs.
Source: Model specification, Pricing, Release notes
- Text
- Image
- Text
Pricing
Every billing leg from the live catalog, in USD per 1M tokens.
| Billing leg | Standard | Long context | Batch |
|---|---|---|---|
| Input | $8 / M | $16 / M | $4 / M |
| Output | $40 / M | $60 / M | $20 / M |
| Cache read | $0.4 / M | $0.8 / M | $0.2 / M |
Long-context rates apply above 272,000 input tokens.
Cost calculator
Billing-grade math over the rate card above, including the thinking tokens most estimates miss.
Hidden reasoning tokens bill as output. 1.0x = thinking off. Medium-effort reasoning typically lands near 3x; heavy reasoning runs 6x to 8x. Set it to match your workload.
Provider
Where this model runs, and what it charges.
| Platform | Input / M | Output / M | Cache read / M | Batch | Access |
|---|---|---|---|---|---|
OpenAI | $8 | $40 | $0.4 | ✓ | First-party API |
Price history
Has this model gotten cheaper or more expensive?
| Effective | Input / Output | Status |
|---|---|---|
| November 17, 2026 – March 1, 2027 | $8 / $40 | current |
| March 1, 2027 | $8 / $40 | upcoming |
More from OpenAI
Other models from the same provider.
Stop estimating. Track it.
SuperPenguin meters every GPT-6.2 Astra call your team makes, prices it with this exact rate card, and shows spend by feature, customer, and team.
Rates are generated from SuperPenguin's live pricing catalog.