Media

Text is only the start. Images, video and speech run through the same key and the same credit pool, priced from exact reported costs plus the flat 20%.

Images

POST /api/v1/images/generations — OpenAI-images shape in:

curl https://sonnely.com/api/v1/images/generations \
  -H "Authorization: Bearer $SONNELY_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"bytedance-seed/seedream-5-0-pro","prompt":"a colosseum at dawn"}'

Featured: Seedream 5 Pro, FLUX 2 Pro, GPT Image 2, Gemini Flash Image. Billing is per render from the exact reported cost — failed renders are free. Full list: GET /api/v1/images/models.

Video

Async by nature: submit, poll, download.

# 1 · submit
curl https://sonnely.com/api/v1/videos/generations \
  -H "Authorization: Bearer $SONNELY_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"google/veo-3.1","prompt":"slow push-in on a neon arena","duration":5}'
# {"id":"...","polling_url":"...","sonnely":{"credits_estimate":...}}

# 2 · poll until completed
curl https://sonnely.com/api/v1/videos/<jobId> \
  -H "Authorization: Bearer $SONNELY_KEY"

Featured: Veo 3.1, Sora 2 Pro, Kling 3 Standard, Hailuo 3 Max. Nothing is charged at submit — the completion poll charges the exact cost once, and failed jobs are never billed. Full list: GET /api/v1/videos/models.

Audio

• POST /api/v1/audio/speech — text to speech, OpenAI-audio shape, raw bytes back. Billed per input character.
• POST /api/v1/audio/transcriptions — multipart audio in, JSON text out. Beta flat estimate per request.

Try it visually

The playground has image and video tabs wired to these same endpoints — same key, same billing.