Arabic AI video: Luma's Arabic UI vs Sume prompts

Luma's September 2, 2026 news adds a native Arabic-language interface for Saudi and Arabic-speaking creators, extending Arabic support across the full Luma Agents interface. Sume's changelog lists localized chrome for Korean (ko-KR) in the Dashboard and Agents, and nothing for Arabic, so on Sume you work in an English or Korean interface and verify Arabic output yourself.

What each side documents

Read the Luma news page for the announcement itself. The interface language and the language of the generated video are separate questions, and the Luma note speaks to the interface.

Verify Arabic output instead of assuming it

Sume's docs I read do not publish a per-model language list for generated speech or on-screen text, so treat Arabic as something to test. Generate a short clip, then run video inspect with transcribe set to true. The language_code field is an STT hint, and leaving it out lets the service auto-detect; the transcript gives you words you can read back against your script.

Check on-screen text visually as well. Right-to-left lettering and joined glyphs are easy to get wrong in generated frames, and a transcript will not catch them. Pull stills with frames and look at them.

curl -X POST https://api.sume.com/v1/video-inspect \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: arabic-check-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/clip.mp4",
    "frames": { "at": [1, 3, 5] },
    "transcribe": true
  }'

A practical routine

• Keep the script in a file and compare the transcript against it line by line.
• Use a native reader for the final check; automated transcripts are a screen, not a sign-off.
• Transcription is billed at $0.01 per audio minute, per the docs; confirm live in the catalog.
• If a model fails the test, change the model before you change the prompt.