Skip to main content
POST
Starts rendering one image per row in rows, all from the same image template. Returns immediately with a bulk_job_id — track progress via Get image batch status or a webhook.
Discovery is shared with video — list image templates and read their schema through List templates and Get template schema with ?type=IMAGE_TEMPLATE. Only generation and results have their own paths: /image-templates/{template_id}/bulk-generate/ and /image-bulk-jobs/. The request body is identical to Generate videos — for a single image, call it with one row in rows.
​
string
required
​
string
required
Bearer {api_key}

​
Body

​
object[]
required
A non-empty list of {variable_name: value} objects, one per image, up to 20,000 per request. Keys should match Get template schema’s variable names — a row missing a key just leaves that placeholder blank in that row’s image.external_id is a reserved key you can include in any row — your own identifier for that row, echoed back unchanged in the CSV export and usable with Get single image status to check a row before the batch finishes. Not treated as a template variable.
​
string
Optional label for the batch, shown back in status responses.
rows — not a CSV file or URL — is the only way to submit a batch through this API. Serialize your data directly into the request body.

​
Errors

​
Response fields

​
string
Use with Get image batch status to track the whole batch.
​
integer
Number of images this batch will render — equal to rows.length.
​
string
0.00181 × variable_count × row_count, deducted upfront. See How images are priced below.

​
How images are priced

Images are priced per variable, not per render — a flat 0.00181 credits for every variable the template declares, charged on every row.
variable_count is the number of variables the template declares — the same set of columns Get template schema returns — not the number of keys a given row happens to fill. Every row in a batch is therefore priced identically, and the count is frozen from the template as it existed when you submitted, so editing the template mid-batch cannot change what you are charged.
This is a different model from video, which charges the template’s own estimated_credits per row. An image batch is typically orders of magnitude cheaper — a 4-variable template over 500 rows costs 0.00181 × 4 × 500 = 3.62 credits.

​
Getting the results

The status endpoints are shared with video and need no change — a batch of images reports through exactly the same Get image batch status, Get single image status and CSV export you already use. One difference in the payload: each row carries output_type: "image", and the finished asset lands in result_image_url. Image rows do not carry result_video_url at all — the field is absent, not null.
Row from an image batch
If you already parse batch results for video, branch on output_type and read result_image_url on image rows. Code that reaches for result_video_url unconditionally will find the key missing rather than empty.
In the CSV export, an image batch’s final column is result_image_url in place of result_video_url — the rest of the header (external_id, generation_id, then the batch’s variable names) is unchanged.