OpenAI First-PartyVerified 2026-11-12
GPT-6 Astra API Pricing & Cost Rules
API Model ID: gpt-6-astra · Context Window: 1,050,000 tokens (Max Input: 922,000) · Max Output: 128,000 tokens
Context Tiers & Token Rate Card (USD per 1M Tokens)
| Tier | Std Input | Std Output | Cache Read | Write (Short) | Write (Long) |
|---|---|---|---|---|---|
| Standard (<= 272K) | $10.00 | $50.00 | $1.00 | $12.5 | $25.00 |
| Long Context (> 272K) | $20.00 | $75.00 | $2.00 | $25.00 | $50.00 |
Prompt Caching Rules
Supported: Yes (automatic)
Minimum Cacheable Prefix: 1,024 tokens
Hourly Cache Storage: $0
Read = 0.10x (90% off). Cache write is a total rate (not additive): 1.25x for 30m in-memory retention, 2.50x for 12h Responses API retention. Min 1,024 visible input tokens.
Batch API & Hosting Notes
Batch Supported: Yes (50% discount · 24h)
Caching in Batch: Allowed + Batch discount stacks onto cache reads
50% off all columns (standard input, output, cached read, 30m write, and 12h write) across short and >272K long context.
Regional data residency & FedRAMP endpoints add +10% uplift on models released >= 2026-03-05. Fast tier = 2x, Ultrafast = 6x.