Sume API output file formats: MP4, PNG, WebP, WAV, MP3
The file each Sume API endpoint returns: MP4 from Timeline and the editing tools, PNG, JPEG, or WebP images, WAV or MP3 audio, and JSON transcripts.
Completed Sume jobs return their files as artifacts on media.sume.com. Timeline, compose, trim, filter, and avatar videos are MP4, and captioned videos are MP4 in current code; image endpoints return PNG, JPEG, or WebP (SVG where a model's catalog allows it); audio detach and Timeline audio return WAV or MP3; and speech-to-text returns JSON with word timings. Where the docs name no container, read the artifact's content_type.
Formats and defaults come from each tool's docs page and the OpenAPI reference, read on 2026-09-27. Encoder settings come from Sume's media compiler code and describe current behavior. The general artifact shape is covered in Video API media inputs and outputs.
What video format does each endpoint return?
In current code, Timeline, compose, and filter encode with libx264, an H.264 encoder, in yuv420p with AAC audio; the trim docs name the same encoder and pixel format for exact trims. For generated clips, lip sync, and upscales, the docs give resolution options but name no container.
| Endpoint | Output | Size, rate, and codec |
|---|---|---|
| Timeline render | MP4 | 1080×1920 by default; frame rate follows the sources unless you set output.fps. Current code: CRF 20, AAC at 192 kbps. |
| Timeline compose | MP4 | 1080×1920 by default, at the video layer's frame rate. Audio passes through from the video. |
| Video trim | MP4 | exact (default) re-encodes frame-accurately; kept audio becomes AAC. keyframe is a stream copy. |
| Video filter | MP4 | Keeps the source's geometry, frame rate, and audio unless the program changes them. Current code: CRF 20, AAC at 192 kbps. |
| Captions | MP4 (current code) | The captioned video_url; the docs name no container. |
| Talking avatar video | MP4 | 720p; aspect_ratio 9:16 by default, or 1:1, 3:4, 4:3, 16:9. |
| Lip sync | Video | VEED Fabric 1.0: 480p or 720p (default 720p). MiniMax H3 Max Lip Sync: 480p, 768p, or 1080p (default 768p). |
Video generation: POST /v1/videos | Video | The resolution and aspect_ratio you request from the model's lists; download from GET /v1/videos/{id}/content with your API key. |
| Video upscale | Video | The input scaled by scale_ratio, 1.1–4, default 2. |
Which image formats can I get?
POST /v1/images takes output_format, but a model only accepts the values its catalog entry lists, so check GET /v1/images/models first. Each result carries media_type, and its data[].url is Sume-hosted and signed.
| Endpoint | Formats | Size |
|---|---|---|
| Image API | png, jpeg, webp, or svg | Set by resolution and aspect_ratio |
| Image upscale | png (default), jpg, or webp | The input times upscale_factor, 1–4, default 2 |
| Background removal | PNG with alpha | — |
| Video frames | jpeg (default) or png | Source size, unless max_edge (16–2160) clamps the long edge |
| Video inspect stills | jpeg (default) or png | max_edge 64–2160, default 768 |
Which audio formats do the audio endpoints return?
Keep WAV when a file will be joined again or will drive lip sync: Timeline audio's MP3 option re-adds priming padding at every edge.
| Endpoint | Formats | Options |
|---|---|---|
| Audio detach | wav (default, pcm_s16le) or mp3 at 128 kbps | sample_rate 16000, 44100, or 48000, or the source's; channels source or mono |
| Timeline audio | wav (default, pcm_s16le) or mp3 | Produced audio up to 1800 seconds |
| Text to speech | mp3 by default at 44,100 Hz and 128 kbps; wav or raw | sample_rate 8000, 16000, 22050, 24000, 44100, or 48000; wav and raw take encoding pcm_f32le, pcm_s16le, pcm_mulaw, or pcm_alaw |
| Music Router | An artifact with type audio | The docs don't name the container |
What format are transcripts in?
JSON with word timings. Speech-to-text returns text and words[] as { word, start, end } in seconds from the audio start. A video inspect with transcribe: true adds a transcript with text, words[], optional sentence segments[], and an audio_url.
How do I check the format of a file I got back?
Read the metadata instead of guessing from the URL. Job artifacts carry content_type, such as image/png, next to their media.sume.com URL, and POST /v1/images results carry media_type. To conform a clip's size or frame rate on the way out, trim's output field and Timeline's output object do it, as changing frame rate or resolution shows.
Sources
Related posts
More in Developers
- Sume API pagination: cursor, starting_after, and page limits
How each Sume list endpoint pages: cursor and has_more on Formats and runs, starting_after on /v1/jobs, and limit-only lists that have no cursor.
- Sume API status values: jobs, runs, queues, and webhooks
Sume API status values in one place: jobs, /v1/videos, Format and Agent runs, bulk queues, webhook deliveries, usage rows, grants, and balance.
- Sume job types and concurrency: which calls take a slot
Each Sume endpoint's job type and slot use: every generation job, trims and Timeline included, takes a concurrency slot; frames and inspect don't.
- Sume API media URL rules: which URLs each endpoint accepts
Sume generation endpoints fetch public HTTPS media URLs. Trim, filter, frames, inspect, and Timeline take only your workspace's media.sume.com URLs.
Written by Sume