Synthesia CSV bulk personalization vs Sume bulk runs (1-100)

Synthesia's bulk personalization takes an uploaded .csv or .xlsx file and maps its columns to template variables. Sume has no spreadsheet upload route: a bulk run is a JSON request with 1 to 100 items, each the same body as a single Format run, executed with a concurrency window of 1 to 16. You convert the rows to JSON yourself.

Synthesia details are from its Bulk personalization page (updated 2026-08-24); Sume details from Bulk runs, read 2026-10-01.

How does Synthesia's flow work?

You pick a template with variables, upload a spreadsheet whose first row is the header, and map columns to variables; videos cannot be generated unless every variable has a column. It then warns that you cannot save your progress before generation: leaving the process loses it.

How does a Sume bulk run work?

POST /v1/formats/{format_id}/bulk-runs (or the {handle}/{slug} twin) queues the items on the server, so you do not drive fan-out from a laptop. items is required and holds 1 to 100 entries in order; concurrency is required and holds an integer 1 to 16, the number of child runs in flight at once. Poll GET /v1/format-run-queues/{id} and read each child at GET /v1/format-runs/{run_id}. The key needs formats:write to create and formats:read to poll.

Can I cancel a bulk queue?

There is no public list-queues or cancel-queue endpoint. Cancel an individual child with POST /v1/format-runs/{run_id}/cancel. Plan for that when sizing a batch, and see Sheets to avatar videos with n8n for a spreadsheet-driven setup.

What should I do with a sheet of more than 100 rows?

Split it into several requests of up to 100 items, and give each its own Idempotency-Key header, as for any submit. The docs do not describe a larger single batch.