Batch processing
Also called Message Batches API, Batches API, batch API
Sending many model requests as one background job, at half the usual price, with results back within 24 hours.
Normally each request to a model gets its answer straight away. A batch is the other arrangement: you send many requests together, they're processed in the background, and you collect the results later. On the Claude API, that costs half the usual price.
The catch is time. Results are ready when the whole batch finishes or after 24 hours, whichever comes first, so batching suits work nobody is waiting for. Results can also arrive out of order, so each request carries an ID you choose, and that ID is how you pair every answer with its request.
For example
A nightly job sends 300 test-generation requests as one batch; by morning the results are in, each matched to its module by a custom_id.
Put it to use
Where you'll meet batch processing in ChatGPT, Gemini and Claude, how to use it there, and the mix-up most people make.
Daily Bytes Plus is in the Android app.