TTS 1.0 rejects a model field: use the TTS router to pick sonic-3.6

TTS 1.0 on Sume rejects a model field, so to choose an engine such as sonic-3.6 you call the TTS Router at /v1/tts-router/generate instead. Both are billed at $0.0475 per 1,000 characters, so a 1,000-character script is $0.0475 either way.

Which route takes which fields

The router request schema is the TTS 1.0 body plus an optional model.

Router request

Everything else is the same body you already use. Fetch the catalog before pinning an id.

curl https://api.sume.com/v1/tts-router/models \
  -H "Authorization: Bearer $SUME_API_KEY"

curl -X POST https://api.sume.com/v1/tts-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: router-demo-01" \
  -d '{"model": "sonic-3.6", "transcript": "Hello from the router.",
       "avatar_handle": "@narrator", "language": "en"}'

Gotchas

sonic-latest and sonic-preview are moving targets; pin sonic-3.6 if you need repeatable output. Unknown ids fail with a model error that points at the catalog. Check the catalog response for per-model constraints before you assume a feature works on every engine.