FLUX Video Edit [fast] is Black Forest Labs’ prompt-guided workflow for changing an existing video. It receives one source clip and an instruction describing the edit, then returns an edited MP4. The source establishes the shot: its duration, framing, camera movement and timing provide the starting structure. Use the prompt to identify what should become different, rather than writing a completely new scene description that leaves the intended change unclear.
The documented edits include removing, adding or replacing objects and characters, rebuilding the setting, restyling a clip, changing visible text and revising dialogue or sound. Start with the target and the operation. An original instruction could remove a bag from a chair while retaining the seated person and the room, or replace a plain mug with a blue enamel mug without changing the hand’s action. If several similar objects are visible, distinguish the target by its position, appearance or role.
For an addition, describe location, scale and relevant action. A tall red lamp beside the sofa is more specific than an unspecified extra lamp somewhere in the shot. For a replacement, identify what becomes what and the movement or contact that should remain. The maker’s prompting guide recommends the shortest instruction that fully names the change, adding detail where the edit requires it. Preserve source features deliberately instead of asking for a general improvement whose scope cannot be evaluated.
A setting change can rebuild the visible environment while retaining the action. The producer demonstrates replacing a climbing surface with a different structure and specifying the new contact points, so the character’s hands and feet still have a meaningful surface. In a practical scene, describe the changed surroundings and the geometry the existing movement relies on. A new background is easier to assess when the original action has an explicit relationship with it.
The guide also shows transforming a rough animation into a finished visual treatment. Name which blockout objects become the intended vehicles or props, then retain their paths, speeds and camera relationship. For restyling, describe the resulting materials, surfaces, light and reflections as well as the aesthetic. A clay-like scene and a photographic-looking scene need different visible cues even when they follow the same action.
A further revision can use the first edited result as its new source. Keep the original and each selected stage, and give the next pass one clear purpose. The documented example follows a rough preview through a finished setting and then another lighting treatment. After each pass, compare the revised clip with the previous source and check that the features you meant to retain are still present.
The workflow carries source audio through unless the instruction requests a dialogue or sound change. A silent source returns a silent result, so it supplies no existing speech to revise. For a dialogue edit, quote the intended line, identify the speaker and fit the new words to the time occupied by the source line. The guide reports a longer replacement being cut when it exceeded the available speech interval. Writing more words does not create extra speaking time inside the existing shot.
Review the audible line alongside the visible performance. Keep a copy of the source wording and intended replacement so an omitted word, mistimed phrase or unwanted sound change can be identified. The useful result is the revised spoken moment inside the edited scene, not an assumed faithful transcript merely because the render completed.
The exact producer profile requires a video and an edit prompt. The instruction is one to 4,096 characters after surrounding whitespace is removed. The clip is supplied through a URL or encoded MP4, with a maximum length of fifteen seconds and a fifty-MiB input limit. Each side must be at least 160 pixels, and the normalized clip must contain at least seventeen frames at 24 fps, roughly seven-tenths of a second. Overlength sources are rejected rather than silently trimmed; undersized sources can fail during processing.
The documented flow uses 24-fps processing and follows the source length and aspect ratio. Inputs above the 720p profile are downscaled; the maker gives a 1920×1088 source returning as 1248×704. Duration, resolution and aspect ratio are not independent output settings in this editing profile. Its sampling recipe is handled internally, and fields such as mode, version, seed or generate_audio are not accepted in the stated request contract.
A video used only as motion or style inspiration for a newly generated shot is a different workflow. This editor does not accept an image reference, audio-only input, masks or a request to extend the clip. Prepare the actual video to alter and retain its selected duration before writing the edit. After asynchronous processing, the producer returns the edited MP4 through a temporary signed delivery link, valid for about one hour. Save the result and compare the changed object, surrounding frames, timing and audio with the source before using the clip in a longer edit.
", "aboutText": "FLUX Video Edit [fast] is Black Forest Labs’ prompt-guided workflow for changing an existing video. It receives one source clip and an instruction describing the edit, then returns an edited MP4. The source establishes the shot: its duration, framing, camera movement and timing provide the starting structure. Use the prompt to identify what should become different, rather than writing a completely new scene description that leaves the intended change unclear.\n\n### Name the change and its position\n\nThe documented edits include removing, adding or replacing objects and characters, rebuilding the setting, restyling a clip, changing visible text and revising dialogue or sound. Start with the target and the operation. An original instruction could remove a bag from a chair while retaining the seated person and the room, or replace a plain mug with a blue enamel mug without changing the hand’s action. If several similar objects are visible, distinguish the target by its position, appearance or role.\n\nFor an addition, describe location, scale and relevant action. A tall red lamp beside the sofa is more specific than an unspecified extra lamp somewhere in the shot. For a replacement, identify what becomes what and the movement or contact that should remain. The maker’s prompting guide recommends the shortest instruction that fully names the change, adding detail where the edit requires it. Preserve source features deliberately instead of asking for a general improvement whose scope cannot be evaluated.\n\n### Settings, styles and rough animation\n\nA setting change can rebuild the visible environment while retaining the action. The producer demonstrates replacing a climbing surface with a different structure and specifying the new contact points, so the character’s hands and feet still have a meaningful surface. In a practical scene, describe the changed surroundings and the geometry the existing movement relies on. A new background is easier to assess when the original action has an explicit relationship with it.\n\nThe guide also shows transforming a rough animation into a finished visual treatment. Name which blockout objects become the intended vehicles or props, then retain their paths, speeds and camera relationship. For restyling, describe the resulting materials, surfaces, light and reflections as well as the aesthetic. A clay-like scene and a photographic-looking scene need different visible cues even when they follow the same action.\n\nA further revision can use the first edited result as its new source. Keep the original and each selected stage, and give the next pass one clear purpose. The documented example follows a rough preview through a finished setting and then another lighting treatment. After each pass, compare the revised clip with the previous source and check that the features you meant to retain are still present.\n\n### Dialogue must fit existing time\n\nThe workflow carries source audio through unless the instruction requests a dialogue or sound change. A silent source returns a silent result, so it supplies no existing speech to revise. For a dialogue edit, quote the intended line, identify the speaker and fit the new words to the time occupied by the source line. The guide reports a longer replacement being cut when it exceeded the available speech interval. Writing more words does not create extra speaking time inside the existing shot.\n\nReview the audible line alongside the visible performance. Keep a copy of the source wording and intended replacement so an omitted word, mistimed phrase or unwanted sound change can be identified. The useful result is the revised spoken moment inside the edited scene, not an assumed faithful transcript merely because the render completed.\n\n### Source limits and inherited output\n\nThe exact producer profile requires a video and an edit prompt. The instruction is one to 4,096 characters after surrounding whitespace is removed. The clip is supplied through a URL or encoded MP4, with a maximum length of fifteen seconds and a fifty-MiB input limit. Each side must be at least 160 pixels, and the normalized clip must contain at least seventeen frames at 24 fps, roughly seven-tenths of a second. Overlength sources are rejected rather than silently trimmed; undersized sources can fail during processing.\n\nThe documented flow uses 24-fps processing and follows the source length and aspect ratio. Inputs above the 720p profile are downscaled; the maker gives a 1920×1088 source returning as 1248×704. Duration, resolution and aspect ratio are not independent output settings in this editing profile. Its sampling recipe is handled internally, and fields such as mode, version, seed or generate_audio are not accepted in the stated request contract.\n\nA video used only as motion or style inspiration for a newly generated shot is a different workflow. This editor does not accept an image reference, audio-only input, masks or a request to extend the clip. Prepare the actual video to alter and retain its selected duration before writing the edit. After asynchronous processing, the producer returns the edited MP4 through a temporary signed delivery link, valid for about one hour. Save the result and compare the changed object, surrounding frames, timing and audio with the source before using the clip in a longer edit.", "editorialOverlay": { "copy_fields": null, "copy_sha256": null, "copy_source": null, "html_sha256": "c4be0889892e1b66bdc870e245e9bc03c8631fabb9457f39a869a94e2f9b1c56", "module_sha256": "ee9dc1c50b1e6a8fb7fa062030fcb4646b9aaaeb08be513b723a5a92a7290f06", "raw_model_unchanged": true, "schema": "ailoot-native-model-editorial-v1" }, "locale": "en", "methods": [ { "html": "black-forest-labs/flux-video-editFLUX Video Edit [fast] transforms an existing clip through a written edit instruction. Add, remove or replace elements, change the setting or style, and revise existing dialogue while working within the documented source-length, framing and audio constraints.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
FLUX Video Edit [fast] is Black Forest Labs’ prompt-guided workflow for changing an existing video. It receives one source clip and an instruction describing the edit, then returns an edited MP4. The source establishes the shot: its duration, framing, camera movement and timing provide the starting structure. Use the prompt to identify what should become different, rather than writing a completely new scene description that leaves the intended change unclear.
The documented edits include removing, adding or replacing objects and characters, rebuilding the setting, restyling a clip, changing visible text and revising dialogue or sound. Start with the target and the operation. An original instruction could remove a bag from a chair while retaining the seated person and the room, or replace a plain mug with a blue enamel mug without changing the hand’s action. If several similar objects are visible, distinguish the target by its position, appearance or role.
For an addition, describe location, scale and relevant action. A tall red lamp beside the sofa is more specific than an unspecified extra lamp somewhere in the shot. For a replacement, identify what becomes what and the movement or contact that should remain. The maker’s prompting guide recommends the shortest instruction that fully names the change, adding detail where the edit requires it. Preserve source features deliberately instead of asking for a general improvement whose scope cannot be evaluated.
A setting change can rebuild the visible environment while retaining the action. The producer demonstrates replacing a climbing surface with a different structure and specifying the new contact points, so the character’s hands and feet still have a meaningful surface. In a practical scene, describe the changed surroundings and the geometry the existing movement relies on. A new background is easier to assess when the original action has an explicit relationship with it.
The guide also shows transforming a rough animation into a finished visual treatment. Name which blockout objects become the intended vehicles or props, then retain their paths, speeds and camera relationship. For restyling, describe the resulting materials, surfaces, light and reflections as well as the aesthetic. A clay-like scene and a photographic-looking scene need different visible cues even when they follow the same action.
A further revision can use the first edited result as its new source. Keep the original and each selected stage, and give the next pass one clear purpose. The documented example follows a rough preview through a finished setting and then another lighting treatment. After each pass, compare the revised clip with the previous source and check that the features you meant to retain are still present.
The workflow carries source audio through unless the instruction requests a dialogue or sound change. A silent source returns a silent result, so it supplies no existing speech to revise. For a dialogue edit, quote the intended line, identify the speaker and fit the new words to the time occupied by the source line. The guide reports a longer replacement being cut when it exceeded the available speech interval. Writing more words does not create extra speaking time inside the existing shot.
Review the audible line alongside the visible performance. Keep a copy of the source wording and intended replacement so an omitted word, mistimed phrase or unwanted sound change can be identified. The useful result is the revised spoken moment inside the edited scene, not an assumed faithful transcript merely because the render completed.
The exact producer profile requires a video and an edit prompt. The instruction is one to 4,096 characters after surrounding whitespace is removed. The clip is supplied through a URL or encoded MP4, with a maximum length of fifteen seconds and a fifty-MiB input limit. Each side must be at least 160 pixels, and the normalized clip must contain at least seventeen frames at 24 fps, roughly seven-tenths of a second. Overlength sources are rejected rather than silently trimmed; undersized sources can fail during processing.
The documented flow uses 24-fps processing and follows the source length and aspect ratio. Inputs above the 720p profile are downscaled; the maker gives a 1920×1088 source returning as 1248×704. Duration, resolution and aspect ratio are not independent output settings in this editing profile. Its sampling recipe is handled internally, and fields such as mode, version, seed or generate_audio are not accepted in the stated request contract.
A video used only as motion or style inspiration for a newly generated shot is a different workflow. This editor does not accept an image reference, audio-only input, masks or a request to extend the clip. Prepare the actual video to alter and retain its selected duration before writing the edit. After asynchronous processing, the producer returns the edited MP4 through a temporary signed delivery link, valid for about one hour. Save the result and compare the changed object, surrounding frames, timing and audio with the source before using the clip in a longer edit.
POST /v1/videostext, image, audio, videovideo| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits black-forest-labs/flux-video-edit | Default— |
Fieldinput_references | Typearray | RequirementRequired | Values & limits Minimum items: 1 Maximum items: 1 | Default— |
Fieldprompt | Typestring | RequirementRequired | Values & limits Minimum length: 1 | Default— |
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, input_references, prompt |
Parametermaterial_types | Contracttext, image, audio, video |
Parametermaterial_conditions | ContractUse exactly one actual completed owned video asset content URL with its returned signature. The asset owner, type and stored byte witness must match the request. |
Lifecycle: asynchronous_job_then_owned_outputs
GET /v1/videos/{request_id} · actual returned id or polling_url
GET /v1/videos/{request_id}/content
The displayed rates estimate the cost before generation. The final charge is calculated from the cost recorded for the original generation. Measured output duration and forecast quantities are not the final billable quantity.
Tariff references do not add request options. Supported parameters are listed for the selected method.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Forecast quantity | Billing unit | Billing conditions | Charge |
|---|---|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Forecast quantity— | Billing unit— | Billing conditions— | Charge Input · Input tokens · Estimate for comparison only — / 1K tokens Output · Output tokens · Estimate for comparison only — / 1K tokens Input · Cache read · Estimate for comparison only — / 1K tokens Input · Cache write · Estimate for comparison only — / 1K tokens Input · Cache write, 1 hour · Estimate for comparison only — / 1K tokens |
| ConfigurationVideo price | Rate schedule— | Forecast quantity— | Billing unitper video second | Billing conditionsNo additional condition is declared in this pricing key. | Charge Output · Video price · Estimate for comparison only $0.06 per video second |
| ConfigurationEstimated generation cost | Rate schedule— | Forecast quantityone generation estimate | Billing unit— | Billing conditions— | Charge Output · Estimated generation cost · Estimate for comparison only $0.06 estimate before generation This estimate is not an additional charge or a fixed rate for the completed generation. |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_INPUT_REFERENCES_JSON:?Set the actual documented input value}"\nexport AILOOT_INPUT_REFERENCES_JSON\n: "${AILOOT_PROMPT:?Set the actual documented input value}"\nexport AILOOT_PROMPT\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\nrequest={'model': 'black-forest-labs/flux-video-edit', 'input_references': '${ACTUAL_OWNED_INPUT_REFERENCES}', 'prompt': '${ACTUAL_EDIT_PROMPT}'}\nrequest['input_references']=json.loads(os.environ['AILOOT_INPUT_REFERENCES_JSON'])\nrequest['prompt']=os.environ['AILOOT_PROMPT']\nPath('request.json').write_text(json.dumps(request,ensure_ascii=False),encoding='utf-8')\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/videos \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o response.bin\n# Retain the complete response and headers, every actual output, original returned identifiers and all usage fields. Recover the original operation without submitting creation again.\nGET /v1/videos/{request_id}GET /v1/videos/{request_id}/contentset -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_STATUS_FILE:?Set the actual completed status JSON file}"\nexport AILOOT_STATUS_FILE\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\nfrom urllib.parse import urlsplit,urljoin\nf=Path(os.environ["AILOOT_STATUS_FILE"])\nif not f.is_file():raise SystemExit("Choose the actual completed status file")\nr=json.loads(f.read_text())\nif r.get("status")!="completed" or not isinstance(r.get("data"),list) or not r["data"]:raise SystemExit("Wait for a completed status with asset paths")\nurls=[]\nfor x in r["data"]:\n if not isinstance(x,str) or "\\n" in x or "\\r" in x:raise SystemExit("Invalid asset path")\n u=urljoin("https://api.ailoot.ai",x)\n z=urlsplit(u)\n if z.scheme!="https" or z.netloc!="api.ailoot.ai" or z.username or z.password:raise SystemExit("Use the actual Ailoot asset paths from status")\n urls.append(u)\nPath("asset-urls.txt").write_text("\\n".join(urls)+"\\n")\nPY_REQUEST\nindex=0\nwhile IFS= read -r asset_url; do\n curl --fail-with-body --silent --show-error "$asset_url" \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -o "video-${index}.mp4"\n index=$((index + 1))\ndone < asset-urls.txt\nSave every completed video or returned file; retrieve every actual returned asset URL.
Retain the complete response and headers, every actual output, original returned identifiers and all usage fields. Recover the original operation without submitting creation again.
Keep X-Client-Request-ID, X-Ailoot-Task-Id and actual returned id/polling_url. Continue the original job; do not submit creation again.
| Field | Type |
|---|---|
Fieldid | Type |
Fieldstatus | Type |
Fielddata[] | Type |
Fieldoutputs[] | Type |
Ailoot Ultra black-forest-labs/flux-video-edit
Use the method paths shown above.