Recently recovered
batchwatch caught a severely degraded OpenAI gpt-5.6-luna batch outage — it recovered after ~16h, 42d ago. The full record is on /v1/outages, measured end to end.
Half the price, measured wait
Half the price. Here is the real wait at OpenAI.
OpenAI's batch API costs 50% of its synchronous rate - the provider's own published number, on every input and output token. The only thing it does not tell you is the wait, and OpenAI publishes just one figure for that: a 24 hours completion window, the ceiling in its contract, not a measurement of the queue.
We measured it. Across OpenAI's models, half of jobs finish between 30s (gpt-5.6-sol) and 85s (gpt-5.6-luna) - far inside the 24 hours the contract allows. That gap is the point: plan against the measured median, keep the published window as the dependable fallback.
Measured models
30-day window. A dash marks a model still building its baseline: we publish a percentile from 2 measurements up, and holding that line is what makes every figure here trustworthy.
| Model | p50 | p90 | p95 | Jobs | Sources | Confidence | Last measured |
|---|---|---|---|---|---|---|---|
| gpt-5-nano | 74s | 79 min | 6 hours | 3881 | 0 | Confidence: Low | 2026-10-10 |
| gpt-5.6-luna | 85s | 15 min | 36 min | 3247 | 0 | Confidence: Low | 2026-10-10 |
| gpt-5.6-sol | 30s | 17 min | 79 min | 1885 | 0 | Confidence: Low | 2026-10-09 |
Questions this page answers
- How long does the OpenAI batch API take?
- It depends on the model. Across the 3 models we have measured in the last 30 days, the median wait ranges from 30s (gpt-5.6-sol) to 85s (gpt-5.6-luna). The table above gives each model, how many jobs that figure rests on, and how much confidence it carries.
- What does OpenAI itself promise?
- A completion window of 24 hours. That is a contract they wrote, not a measurement - a ceiling everyone quotes because nobody has timed the real thing. We have, and the measured figures above are the ones worth planning against; OpenAI's window is the dependable fallback until a model has the sample to beat it.
- How current are these numbers?
- The newest measurement behind this page finished at 2026-10-10T01:44:32.000Z. Public figures are precomputed and delayed by 15 minutes; anyone contributing measurements reads them live, with no delay at all. Every row states its own age rather than relying on this page being fresh.
This question has its own measured answer page: How long does the OpenAI batch API take?
Get this as JSON
No key required for coverage.
curl "https://batchwatch.dev/v1/coverage"
Per-model curves live at
/v1/curve?provider=openai&model=....
The front page has the decision endpoints.