2h and 8h background windows in Responses API beta: 30% and 40% lower token rates
OpenAI adds 2h and 8h completion windows for background Responses API requests, with model token rates 30% and 40% lower than Standard.
On Zabio today
gpt-6.2-astraLive. Call it by name with your Zabio key.Price and detailsgpt-6.2-solLive. Call it by name with your Zabio key.Price and detailsgpt-6.1-terraLive. Call it by name, or let auto pick it when it fits the job.Price and detailsgpt-6-lunaLive. Call it by name with your Zabio key.Price and details
What changed
OpenAI added a choice of two completion windows for background mode in the Responses API, now in public beta. The 2h window targets completion within two hours after the API accepts the request, and the 8h window targets eight hours. Model token rates are 30% lower than Standard for the 2h window and 40% lower for the 8h window. The default soonest window has no discount.
Key details
- Discounted windows are in public beta on GPT-6.2 Astra, GPT-6.2 Sol, GPT-6.1 Terra, and GPT-6 Luna.
- Requests with 2h or 8h windows share the Batch/Background Rate Limit with Batch API jobs.
- Discounted windows require Standard processing: set service_tier to default.
- The discount applies to each eligible request, including follow-up requests.
- With store set to false or Zero Data Retention, background mode still stores response data temporarily. For 2h and 8h requests, that content is encrypted at rest.
What it means for Zabio customers
If your application can wait for the initial response, a 2h or 8h window lowers token costs compared with the default soonest window. Follow-up requests with the same model, service tier, background setting, completion window, and continuation ID also get the discount.
Source
Choose a two-hour or eight-hour background completion window in the Responses API public beta. Model token rates are 30% lower than Standard for the two-hour window and 40% lower for the eight-hour window.
Original announcement: OpenAI API changelog
How it compares
The models in this story next to models of the same class at a similar price, all available with one Zabio key.
| Model | Input, $/M | Output, $/M | Cached, $/M | Context window | 1,000 calls | Writing code, hard |
|---|---|---|---|---|---|---|
gpt-6.2-astraOpenAIIn this story | 8.00 | 40.00 | 0.40 | 8,192 | $36.00 | no data |
gpt-6.2-solOpenAIIn this story | 2.00 | 10.00 | 0.10 | 8,192 | $9.00 | no data |
gpt-6.1-terraOpenAIIn this story | 1.00 | 5.00 | 0.05 | 8,192 | $4.50 | 66.4 |
gpt-6-lunaOpenAIIn this story | 0.10 | 0.50 | 0.01 | 400,000 | $0.45 | 25.7 |
claude-fable-5-2Anthropic | 8.00 | 40.00 | 0.20 | 1,000,000 | $36.00 | 82.9 |
kimi-k3.1-pro-highspeedMoonshot AI | 4.00 | 20.00 | 0.20 | 1,048,576 | $18.00 | no data |
gemini-4-neonGoogle | 3.00 | 15.00 | no data | 200,000 | $13.50 | no data |
Prices are the vendor list prices in USD per million tokens. "1,000 calls" is 1,000 requests of 2,000 input tokens and 500 output tokens each, without caching. Zabio bills the list price plus its service fee: see pricing.
Scores: percent correct for that kind of job, aggregated from public results by Epoch AI, LMArena, and LiveBench, read on 16 November 2026. Epoch AI: CC BY 4.0, LMArena: CC BY 4.0, LiveBench: Leaderboard results used with attribution; the 2.0 release states no terms of its own (read 2026-11-06); the first version's Mathematics, Language, IF results: Code Apache 2.0 and MIT; leaderboard results used with attribution
Source: OpenAI API changelog (2026-11-17)