Batch APIs let you submit many AI requests at once and get the results later, in exchange for a lower price. OpenAI and Anthropic both charge 50% of standard prices for batch work, with a 24-hour window. Batches suit work that doesn't need an instant answer, such as tagging a backlog or generating descriptions overnight.
Verified Oct 18, 2026Review by Apr 16, 20272 sources
What it is
A batch API accepts a file or list of requests, processes them asynchronously within a stated window, and returns the results together. Vendors price batch work below real-time calls.
How it works
Prepare a file of requests, one per item.
Submit it to the batch endpoint and note the job ID.
Collect results and errors separately. Match each result to its unique request ID rather than its line position, then load only successful results into your system.
Vendor terms (checked 2026-10-18):
OpenAI’s Batch API offers 50% lower costs and a separate limit pool, with a 24-hour completion window. A batch accepts up to 50,000 requests and a 200 MB input file. Unfinished requests can expire; completed requests are still charged.
Anthropic’s Message Batches API charges 50% of standard prices. Most batches finish within an hour, and batches that don’t complete within 24 hours expire. A batch can hold up to 100,000 requests or 256 MB, and results stay available for 29 days from batch creation.
Benefits
Lower cost for bulk work.
Separate batch capacity from real-time rate limits, with its own limits still enforced.
A simple pattern for nightly reporting and enrichment.
Limits and risks
No instant answers. Results arrive later.
Validate the request file and a small sample first. Handle per-item failures and expired requests separately; a completed batch is not proof every item succeeded.
Anthropic notes that batches may go slightly over a workspace’s spend limit because of concurrent processing.
Discounts and windows change, so re-check the vendor page before you plan around them.
Checked specs
Snapshot from official sources on Oct 18, 2026.
OpenAI batch discount and window
50% lower cost; 24-hour turnaround; separate, higher rate-limit pool [1]
Anthropic batch discount and window
50% of standard prices; most batches finish within 1 hour; expire after 24 hours; up to 100,000 requests or 256 MB [2]
Practitioner views are restated in our words and labeled as views, not measured results. Product facts come from official sources dated above.
Suggest an edit
IonBin's wiki isn't user-edited. Corrections go to our Directory Curator, who checks them against primary sources before anything changes. Spot something wrong or out of date on “Batch processing (batch APIs)”? Email us with a link to a primary source.