Release2026-09-22
OpenRouter launched a Batch API for bundling requests. It cuts per-token prices by half across 70 models. Median turnaround in beta was 7 minutes. 90% of batches finished within an hour. Batches support chat, responses, messages and embeddings. Providers choose when to process within 24 hours. Results are retained for 30 days unless deleted. Web search calls bill at standard rates. Images and files must use public URLs. Audio, video and web search are not supported.
What changed
Requests were processed immediately at standard per-token prices.
What it unlocks
Submitting thousands of requests in one call and collecting results within 24 hours.
- 50% cheaper per-token price
- 70+ supported models
- 7 minutes median turnaround
- 90% under 1 hour
What you need to act on it
- accepting variable response times up to 24 hours
Sources