Skip to main content
Glama

var2_create_video

Create video. Replace/swap something in an existing clip, recast it with a new character, or restyle it into a preset look (anime, claymation…) → use var2_magic_replace instead. Pick type by INPUT: existing clip — www.var2.ai/video/ or video link (free-form prompt edit, "keep the motion") → video-to-video (video_url), NEVER image-to-video for a /video/ link. ONE image to animate → image-to-video. 2-3 image links, one consistent character in a NEW scene → reference-to-video (reference_image_urls), not kling_elements. Portrait image + audio track (talking head) → audio-to-video. Text only → text-to-video. "Turn this image into a video" / "animate this image" — link or URL included — is THIS tool, type=image-to-video, URL as first_frame_url (grok-imagine: reference_image_url); an action verb on an image link beats the /image/ path — NOT var2_get_image_result. Video only — songs/music are a different tool. Models: veo-3.1 (default), ltx-2.3, ltx-retake, wan-2.7, seedance-2, seedance-2.5, seedance-2-mini, happyhorse, kling, kling-3, kling-motion-control, kling-3-motion-control, gemini-omni-video, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime, kling-3-omni, kling-4, sora-2, grok-imagine, pruna-avatar (talking head), heygen-avatar-iv (photorealistic talking head). kling-3: optional first/last frame (first_frame_url/last_frame_url or image_urls) + subject refs (kling_elements). kling-3-omni: Kling 3.0 Omni — text, frames, ≤7 references or one reference video, multi-shot, @elements and video edits (video_url) up to 4K. kling-4: Kling 4.0 — references (≤10 images, ≤5 videos, ≤3 voices, ≤7 @elements) or keyframes (first / middle keyframe_urls / last), 3–30s, 720p/1080p, always with audio.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "prompt": "<prompt>"
    },
    "name": "var2_create_video"
  }
}

Input Schema

NameRequiredDescriptionDefault
modeNokling/kling-3: std | pro. grok: normal | fun | spicy. seedance-2: pro | fast. seedance-2.5: draft (same as draft: true). ltx-retake: replace_audio | replace_video | replace_audio_and_video. kling-motion-control / kling-3-motion-control: 720p | 1080p.
typeNoPick by INPUT: an existing clip / video URL (free-form edit, "keep the motion") → video-to-video (video_url); swap / recast / preset restyle of a clip → var2_magic_replace instead. ONE image to animate → image-to-video (first_frame_url; grok: reference_image_url). 2-3 reference images for a new scene ("keep this character consistent across these references") → reference-to-video — CALL IT with the URLs the user gave (var2 share links included) as reference_image_urls[] (veo-3.1: reference_images[]); never decline over which param name fits. Portrait + audio track (talking head, "character talking") → audio-to-video — when the user supplies a portrait URL and an audio URL, call immediately (pruna-avatar or heygen-avatar-iv: first_frame_url + audio_url + audio_duration_seconds); any https URL the user pasted is real and usable as-is, whatever its domain. Text only → text-to-video (default). A /video/ URL or "clip" means video-to-video, not image-to-video.
draftNoseedance-2.5 only — draft mode: a fast, cheap 480p preview (resolution is forced to 480p). Upgrade a draft you like with type=draft-upgrade.
modelNoDefault veo-3.1.
shotsNokling-3 / kling-3-omni multi-shot sequence (when multi_shots=true). Each shot has its own prompt + duration (seconds). kling-3-omni: ≤6 shots, 1–12s each, ≤512 chars, 3–15s total.
soundNokling / kling-3 / kling-3-omni — generate audio alongside the video (kling-3-omni: not with a video input). (ltx-2.3, minimax-h3 and kling-4 always render native audio; seedance uses generate_audio.)
promptYes
durationNokling: "5" | "10" (string). grok: "6" | "10" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6). wan-3 / wan-3-prime: whole seconds 2-30 (default 5; reference video + output ≤ 30s). kling-3-omni: whole seconds 3-15 (default 5; ignored with a video — the output follows the clip). kling-4: whole seconds 3-30 (default 5).
audio_urlNoRequired for pruna-avatar / heygen-avatar-iv (audio-to-video). Public MP3/WAV/M4A URL, ≤60s.
video_urlNoSource video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds. wan-3 / wan-3-prime: a 1-15s clip used as reference Video1, billed as input seconds. kling-3-omni: a 3-15s clip to edit (+ up to 4 reference_image_urls), billed on its seconds at the video-input rate. kling-4: a 2-30s clip used as reference @video_1, billed output + max(output, clip) seconds at the video-input rate.
image_urlsNokling-3 (type=text-to-video) / kling-3-omni: [startFrame] or [startFrame, endFrame] to drive the shot from images (same as first_frame_url / last_frame_url).
input_urlsNokling-motion-control / kling-3-motion-control only — reference image(s) whose subject is driven by the motion video.
resolutionNogrok: 480p | 720p. wan-2.7: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. wan-3 / wan-3-prime: 480p | 720p (default) | 1080p. kling-3-omni: 720p (default) | 1080p | 4k. kling-4: 720p (default) | 1080p. Others: model-specific.
start_timeNoltx-retake only — segment start offset in seconds.
multi_shotsNokling-3 / kling-3-omni — true to render a multi-shot sequence from `shots` (kling-3-omni: not with a video).
webhook_urlNoOptional. https:// URL var2 POSTs a signed delivery to when the asset is ready (or fails). Signature header: X-VAR2-Signature: t=<unix>, v1=<hex(hmac_sha256(webhook_secret, t + "." + body))>. Retried with exponential backoff (8 attempts over ~4 days).
aspect_ratioNokling/grok: 1:1 | 16:9 | 9:16. veo/sora: portrait | landscape. wan-3 / wan-3-prime: auto (default) | 16:9 | 9:16 | 1:1 | 4:3 | 3:4.
keyframe_urlsNokling-4 keyframe mode only — middle frames in chronological order (≤8, and first + middle + last ≤ duration+1, 10 from 9s up), between first_frame_url and last_frame_url. Mention them as @keyframe_1, @keyframe_2….
generate_audioNoseedance-2 / seedance-2.5 / seedance-2-mini / wan-3 / wan-3-prime — generate audio (default true).
kling_elementsNokling-3 (type=text-to-video) / kling-3-omni / kling-4: subject reference images for character/object consistency. Each element is a named subject backed by 2+ images (kling-3-omni: 2–4, ≤3 elements, one-word names, not with a video; kling-4: 2–4 images OR one 3–8s character video_url, ≤7 elements / ≤3 video elements, names 1–20 letters/digits/_, optional tag + voice_url); write @name in the prompt to place it.
last_frame_urlNoOptional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime, kling-3-omni, kling-4 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)
total_durationNokling-3 only — total seconds (5–15). Defaults to the sum of shot durations.
video_durationNoltx-retake: full trimmed source duration (drives per-second pricing). kling-motion-control: reference video duration (≤30s); kling-3-motion-control reads it server-side.
first_frame_urlNoStarting frame for image-to-video (all models except grok-imagine); also accepted by kling-3 (type=text-to-video) and used as the portrait for pruna-avatar / heygen-avatar-iv.
negative_promptNokling / kling-3 / wan-2.7 — things to avoid in the output.
reference_imagesNoveo-3.1 reference-to-video only — 1–3 character/style reference images (veo uses this param, not reference_image_urls).
reference_labelsNoseedance-2 / seedance-2.5 / seedance-2-mini / minimax-h3 / kling-4 — optional short names for the reference media, aligned by index with reference_image_urls / reference_video_urls / reference_audio_urls (e.g. {"images": ["david", "castle at night"], "videos": ["dolly move"]}). The server prepends them to the prompt as a "[Reference assets]" mapping ("@Image 1: david") so the model binds each asset; leave an entry null to use the type default. ≤64 per type, ≤80 chars each.
source_video_urlNoDeprecated alias for video_url (video-to-video). Prefer video_url.
draft_generation_idNoseedance-2.5 type=draft-upgrade only — the placeholder_id (or share link) of a finished seedance-2.5 draft. Renders that exact video at 1080p; prompt, aspect ratio and length come from the draft. Drafts can be upgraded for 7 days.
keep_original_soundNokling-4 with reference videos — true keeps the reference video's own audio instead of generating new sound (default false).
reference_image_urlNogrok-imagine image-to-video only (singular, distinct from reference_image_urls).
reference_audio_urlsNoreference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks (≤15s combined); seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside); wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, free) — name them Audio1, Audio2… in the prompt; kling-4: up to 3 voices (free), mentioned as @voice_1, @voice_2….
reference_image_urlsNoreference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3 / wan-3 / wan-3-prime / kling-3-omni) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee; wan-3 / wan-3-prime up to 10, free — name them Image1, Image2… in the prompt; kling-3-omni up to 7 shared with kling_elements, 4 next to a video; kling-4 up to 10, free — mention them as @image_1, @image_2… (in a kling-4 keyframe job they become the middle keyframes). veo-3.1 reference mode uses reference_images instead.
reference_video_urlsNoseedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output. wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, + output ≤ 30s), billed like minimax-h3 — name them Video1, Video2… in the prompt. kling-3-omni: ONE 3-15s clip (+ up to 4 reference_image_urls); the output follows its length, billed at the video-input rate. kling-4: up to 5 (each 2-30s, ≤30s total incl. video elements), mention them as @video_1…; billed output + max(output, input) seconds at the video-input rate.
character_orientationNokling-motion-control / kling-3-motion-control only — take orientation from the reference image or the motion video.
audio_duration_secondsNoRequired for pruna-avatar / heygen-avatar-iv — drives per-second pricing.
reference_video_durationsNoSeconds per entry of reference_video_urls. Optional — the server measures the files itself for billing; supplying it only makes var2_estimate_cost exact.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed17 schema fields changed
    • changedInput schema / properties / duration / description
      Previous value: -"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6). wan-3 / wan-3-prime: whole seconds 2-30 (default 5; reference video + output ≤ 30s). kling-3-omni: whole seconds 3-15 (default 5; ignored with a video — the output follows the clip)."New value: +"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6). wan-3 / wan-3-prime: whole seconds 2-30 (default 5; reference video + output ≤ 30s). kling-3-omni: whole seconds 3-15 (default 5; ignored with a video — the output follows the clip). kling-4: whole seconds 3-30 (default 5)."
    • addedInput schema / properties / keep_original_sound
      Added value: +{
      +  "description": "kling-4 with reference videos — true keeps the reference video's own audio instead of generating new sound (default false).",
      +  "type": "boolean"
      +}
    • addedInput schema / properties / keyframe_urls
      Added value: +{
      +  "description": "kling-4 keyframe mode only — middle frames in chronological order (≤8, and first + middle + last ≤ duration+1, 10 from 9s up), between first_frame_url and last_frame_url. Mention them as @keyframe_1, @keyframe_2….",
      +  "items": {
      +    "type": "string"
      +  },
      +  "type": "array"
      +}
    • changedInput schema / properties / kling_elements / description
      Previous value: -"kling-3 (type=text-to-video) / kling-3-omni: subject reference images for character/object consistency. Each element is a named subject backed by 2+ images (kling-3-omni: 2–4, ≤3 elements, one-word names, not with a video); write @name in the prompt to place it."New value: +"kling-3 (type=text-to-video) / kling-3-omni / kling-4: subject reference images for character/object consistency. Each element is a named subject backed by 2+ images (kling-3-omni: 2–4, ≤3 elements, one-word names, not with a video; kling-4: 2–4 images OR one 3–8s character video_url, ≤7 elements / ≤3 video elements, names 1–20 letters/digits/_, optional tag + voice_url); write @name in the prompt to place it."
    • addedInput schema / properties / kling_elements / items / properties / tag
      Added value: +{
      +  "description": "kling-4 only — subject category (video elements are always character).",
      +  "enum": [
      +    "character",
      +    "animal",
      +    "prop",
      +    "costume",
      +    "scene",
      +    "effect",
      +    "others"
      +  ],
      +  "type": "string"
      +}
    • addedInput schema / properties / kling_elements / items / properties / video_url
      Added value: +{
      +  "description": "kling-4 only — one 3–8s single-shot character video instead of images (var2-hosted).",
      +  "type": "string"
      +}
    • addedInput schema / properties / kling_elements / items / properties / voice_url
      Added value: +{
      +  "description": "kling-4 only — reference audio bound to a character element.",
      +  "type": "string"
      +}
    • changedInput schema / properties / kling_elements / items / required
      Previous value: -[
      -  "name",
      -  "element_input_urls"
      -]New value: +[
      +  "name"
      +]
    • changedInput schema / properties / last_frame_url / description
      Previous value: -"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime, kling-3-omni (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"New value: +"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime, kling-3-omni, kling-4 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"
    • changedInput schema / properties / model / enum
      Previous value: -[
      -  "veo-3.1",
      -  "ltx-2.3",
      -  "ltx-retake",
      -  "wan-2.7",
      -  "seedance-2",
      -  "seedance-2.5",
      -  "seedance-2-mini",
      -  "happyhorse",
      -  "kling",
      -  "kling-3",
      -  "kling-motion-control",
      -  "kling-3-motion-control",
      -  "gemini-omni-video",
      -  "gemini-omni-flash-1-1",
      -  "minimax-h3",
      -  "wan-3",
      -  "wan-3-prime",
      -  "kling-3-omni",
      -  "sora-2",
      -  "grok-imagine",
      -  "pruna-avatar",
      -  "heygen-avatar-iv"
      -]New value: +[
      +  "veo-3.1",
      +  "ltx-2.3",
      +  "ltx-retake",
      +  "wan-2.7",
      +  "seedance-2",
      +  "seedance-2.5",
      +  "seedance-2-mini",
      +  "happyhorse",
      +  "kling",
      +  "kling-3",
      +  "kling-motion-control",
      +  "kling-3-motion-control",
      +  "gemini-omni-video",
      +  "gemini-omni-flash-1-1",
      +  "minimax-h3",
      +  "wan-3",
      +  "wan-3-prime",
      +  "kling-3-omni",
      +  "kling-4",
      +  "sora-2",
      +  "grok-imagine",
      +  "pruna-avatar",
      +  "heygen-avatar-iv"
      +]
    • changedInput schema / properties / reference_audio_urls / description
      Previous value: -"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks (≤15s combined); seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside); wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, free) — name them Audio1, Audio2… in the prompt."New value: +"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks (≤15s combined); seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside); wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, free) — name them Audio1, Audio2… in the prompt; kling-4: up to 3 voices (free), mentioned as @voice_1, @voice_2…."
    • changedInput schema / properties / reference_image_urls / description
      Previous value: -"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3 / wan-3 / wan-3-prime / kling-3-omni) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee; wan-3 / wan-3-prime up to 10, free — name them Image1, Image2… in the prompt; kling-3-omni up to 7 shared with kling_elements, 4 next to a video). veo-3.1 reference mode uses reference_images instead."New value: +"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3 / wan-3 / wan-3-prime / kling-3-omni) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee; wan-3 / wan-3-prime up to 10, free — name them Image1, Image2… in the prompt; kling-3-omni up to 7 shared with kling_elements, 4 next to a video; kling-4 up to 10, free — mention them as @image_1, @image_2… (in a kling-4 keyframe job they become the middle keyframes). veo-3.1 reference mode uses reference_images instead."
    • changedInput schema / properties / reference_labels / description
      Previous value: -"seedance-2 / seedance-2.5 / seedance-2-mini / minimax-h3 — optional short names for the reference media, aligned by index with reference_image_urls / reference_video_urls / reference_audio_urls (e.g. {\"images\": [\"david\", \"castle at night\"], \"videos\": [\"dolly move\"]}). The server prepends them to the prompt as a \"[Reference assets]\" mapping (\"@Image 1: david\") so the model binds each asset; leave an entry null to use the type default. ≤64 per type, ≤80 chars each."New value: +"seedance-2 / seedance-2.5 / seedance-2-mini / minimax-h3 / kling-4 — optional short names for the reference media, aligned by index with reference_image_urls / reference_video_urls / reference_audio_urls (e.g. {\"images\": [\"david\", \"castle at night\"], \"videos\": [\"dolly move\"]}). The server prepends them to the prompt as a \"[Reference assets]\" mapping (\"@Image 1: david\") so the model binds each asset; leave an entry null to use the type default. ≤64 per type, ≤80 chars each."
    • changedInput schema / properties / reference_video_urls / description
      Previous value: -"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output. wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, + output ≤ 30s), billed like minimax-h3 — name them Video1, Video2… in the prompt. kling-3-omni: ONE 3-15s clip (+ up to 4 reference_image_urls); the output follows its length, billed at the video-input rate."New value: +"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output. wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, + output ≤ 30s), billed like minimax-h3 — name them Video1, Video2… in the prompt. kling-3-omni: ONE 3-15s clip (+ up to 4 reference_image_urls); the output follows its length, billed at the video-input rate. kling-4: up to 5 (each 2-30s, ≤30s total incl. video elements), mention them as @video_1…; billed output + max(output, input) seconds at the video-input rate."
    • changedInput schema / properties / resolution / description
      Previous value: -"grok: 480p | 720p. wan-2.7: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. wan-3 / wan-3-prime: 480p | 720p (default) | 1080p. kling-3-omni: 720p (default) | 1080p | 4k. Others: model-specific."New value: +"grok: 480p | 720p. wan-2.7: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. wan-3 / wan-3-prime: 480p | 720p (default) | 1080p. kling-3-omni: 720p (default) | 1080p | 4k. kling-4: 720p (default) | 1080p. Others: model-specific."
    • changedInput schema / properties / sound / description
      Previous value: -"kling / kling-3 / kling-3-omni — generate audio alongside the video (kling-3-omni: not with a video input). (ltx-2.3 always renders native audio; seedance uses generate_audio.)"New value: +"kling / kling-3 / kling-3-omni — generate audio alongside the video (kling-3-omni: not with a video input). (ltx-2.3, minimax-h3 and kling-4 always render native audio; seedance uses generate_audio.)"
    • changedInput schema / properties / video_url / description
      Previous value: -"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds. wan-3 / wan-3-prime: a 1-15s clip used as reference Video1, billed as input seconds. kling-3-omni: a 3-15s clip to edit (+ up to 4 reference_image_urls), billed on its seconds at the video-input rate."New value: +"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds. wan-3 / wan-3-prime: a 1-15s clip used as reference Video1, billed as input seconds. kling-3-omni: a 3-15s clip to edit (+ up to 4 reference_image_urls), billed on its seconds at the video-input rate. kling-4: a 2-30s clip used as reference @video_1, billed output + max(output, clip) seconds at the video-input rate."
  2. Changed1 schema field changed
    • addedInput schema / properties / reference_labels
      Added value: +{
      +  "additionalProperties": false,
      +  "description": "seedance-2 / seedance-2.5 / seedance-2-mini / minimax-h3 — optional short names for the reference media, aligned by index with reference_image_urls / reference_video_urls / reference_audio_urls (e.g. {\"images\": [\"david\", \"castle at night\"], \"videos\": [\"dolly move\"]}). The server prepends them to the prompt as a \"[Reference assets]\" mapping (\"@Image 1: david\") so the model binds each asset; leave an entry null to use the type default. ≤64 per type, ≤80 chars each.",
      +  "properties": {
      +    "audios": {
      +      "items": {
      +        "maxLength": 80,
      +        "type": [
      +          "string",
      +          "null"
      +        ]
      +      },
      +      "maxItems": 64,
      +      "type": "array"
      +    },
      +    "images": {
      +      "items": {
      +        "maxLength": 80,
      +        "type": [
      +          "string",
      +          "null"
      +        ]
      +      },
      +      "maxItems": 64,
      +      "type": "array"
      +    },
      +    "videos": {
      +      "items": {
      +        "maxLength": 80,
      +        "type": [
      +          "string",
      +          "null"
      +        ]
      +      },
      +      "maxItems": 64,
      +      "type": "array"
      +    }
      +  },
      +  "type": "object"
      +}
  3. Changed12 schema fields changed
    • changedInput schema / properties / duration / description
      Previous value: -"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6). wan-3 / wan-3-prime: whole seconds 2-30 (default 5; reference video + output ≤ 30s)."New value: +"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6). wan-3 / wan-3-prime: whole seconds 2-30 (default 5; reference video + output ≤ 30s). kling-3-omni: whole seconds 3-15 (default 5; ignored with a video — the output follows the clip)."
    • changedInput schema / properties / image_urls / description
      Previous value: -"kling-3 only (type=text-to-video): [startFrame] or [startFrame, endFrame] to drive the shot from images."New value: +"kling-3 (type=text-to-video) / kling-3-omni: [startFrame] or [startFrame, endFrame] to drive the shot from images (same as first_frame_url / last_frame_url)."
    • changedInput schema / properties / kling_elements / description
      Previous value: -"kling-3 only (type=text-to-video): subject reference images for character/object consistency. Each element is a named subject backed by 2+ images."New value: +"kling-3 (type=text-to-video) / kling-3-omni: subject reference images for character/object consistency. Each element is a named subject backed by 2+ images (kling-3-omni: 2–4, ≤3 elements, one-word names, not with a video); write @name in the prompt to place it."
    • changedInput schema / properties / last_frame_url / description
      Previous value: -"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"New value: +"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime, kling-3-omni (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"
    • changedInput schema / properties / model / enum
      Previous value: -[
      -  "veo-3.1",
      -  "ltx-2.3",
      -  "ltx-retake",
      -  "wan-2.7",
      -  "seedance-2",
      -  "seedance-2.5",
      -  "seedance-2-mini",
      -  "happyhorse",
      -  "kling",
      -  "kling-3",
      -  "kling-motion-control",
      -  "kling-3-motion-control",
      -  "gemini-omni-video",
      -  "gemini-omni-flash-1-1",
      -  "minimax-h3",
      -  "wan-3",
      -  "wan-3-prime",
      -  "sora-2",
      -  "grok-imagine",
      -  "pruna-avatar",
      -  "heygen-avatar-iv"
      -]New value: +[
      +  "veo-3.1",
      +  "ltx-2.3",
      +  "ltx-retake",
      +  "wan-2.7",
      +  "seedance-2",
      +  "seedance-2.5",
      +  "seedance-2-mini",
      +  "happyhorse",
      +  "kling",
      +  "kling-3",
      +  "kling-motion-control",
      +  "kling-3-motion-control",
      +  "gemini-omni-video",
      +  "gemini-omni-flash-1-1",
      +  "minimax-h3",
      +  "wan-3",
      +  "wan-3-prime",
      +  "kling-3-omni",
      +  "sora-2",
      +  "grok-imagine",
      +  "pruna-avatar",
      +  "heygen-avatar-iv"
      +]
    • changedInput schema / properties / multi_shots / description
      Previous value: -"kling-3 only — true to render a multi-shot sequence from `shots`."New value: +"kling-3 / kling-3-omni — true to render a multi-shot sequence from `shots` (kling-3-omni: not with a video)."
    • changedInput schema / properties / reference_image_urls / description
      Previous value: -"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3 / wan-3 / wan-3-prime) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee; wan-3 / wan-3-prime up to 10, free — name them Image1, Image2… in the prompt). veo-3.1 reference mode uses reference_images instead."New value: +"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3 / wan-3 / wan-3-prime / kling-3-omni) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee; wan-3 / wan-3-prime up to 10, free — name them Image1, Image2… in the prompt; kling-3-omni up to 7 shared with kling_elements, 4 next to a video). veo-3.1 reference mode uses reference_images instead."
    • changedInput schema / properties / reference_video_urls / description
      Previous value: -"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output. wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, + output ≤ 30s), billed like minimax-h3 — name them Video1, Video2… in the prompt."New value: +"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output. wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, + output ≤ 30s), billed like minimax-h3 — name them Video1, Video2… in the prompt. kling-3-omni: ONE 3-15s clip (+ up to 4 reference_image_urls); the output follows its length, billed at the video-input rate."
    • changedInput schema / properties / resolution / description
      Previous value: -"grok: 480p | 720p. wan-2.7: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. wan-3 / wan-3-prime: 480p | 720p (default) | 1080p. Others: model-specific."New value: +"grok: 480p | 720p. wan-2.7: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. wan-3 / wan-3-prime: 480p | 720p (default) | 1080p. kling-3-omni: 720p (default) | 1080p | 4k. Others: model-specific."
    • changedInput schema / properties / shots / description
      Previous value: -"kling-3 multi-shot sequence (when multi_shots=true). Each shot has its own prompt + duration (seconds)."New value: +"kling-3 / kling-3-omni multi-shot sequence (when multi_shots=true). Each shot has its own prompt + duration (seconds). kling-3-omni: ≤6 shots, 1–12s each, ≤512 chars, 3–15s total."
    • changedInput schema / properties / sound / description
      Previous value: -"kling / kling-3 — generate audio alongside the video. (ltx-2.3 always renders native audio; seedance uses generate_audio.)"New value: +"kling / kling-3 / kling-3-omni — generate audio alongside the video (kling-3-omni: not with a video input). (ltx-2.3 always renders native audio; seedance uses generate_audio.)"
    • changedInput schema / properties / video_url / description
      Previous value: -"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds. wan-3 / wan-3-prime: a 1-15s clip used as reference Video1, billed as input seconds."New value: +"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds. wan-3 / wan-3-prime: a 1-15s clip used as reference Video1, billed as input seconds. kling-3-omni: a 3-15s clip to edit (+ up to 4 reference_image_urls), billed on its seconds at the video-input rate."
  4. Changed10 schema fields changed
    • changedInput schema / properties / aspect_ratio / description
      Previous value: -"kling/grok: 1:1 | 16:9 | 9:16. veo/sora: portrait | landscape."New value: +"kling/grok: 1:1 | 16:9 | 9:16. veo/sora: portrait | landscape. wan-3 / wan-3-prime: auto (default) | 16:9 | 9:16 | 1:1 | 4:3 | 3:4."
    • changedInput schema / properties / duration / description
      Previous value: -"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6)."New value: +"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6). wan-3 / wan-3-prime: whole seconds 2-30 (default 5; reference video + output ≤ 30s)."
    • changedInput schema / properties / generate_audio / description
      Previous value: -"seedance-2 / seedance-2.5 / seedance-2-mini — generate audio (default true)."New value: +"seedance-2 / seedance-2.5 / seedance-2-mini / wan-3 / wan-3-prime — generate audio (default true)."
    • changedInput schema / properties / last_frame_url / description
      Previous value: -"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"New value: +"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3, wan-3, wan-3-prime (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"
    • changedInput schema / properties / model / enum
      Previous value: -[
      -  "veo-3.1",
      -  "ltx-2.3",
      -  "ltx-retake",
      -  "wan-2.7",
      -  "seedance-2",
      -  "seedance-2.5",
      -  "seedance-2-mini",
      -  "happyhorse",
      -  "kling",
      -  "kling-3",
      -  "kling-motion-control",
      -  "kling-3-motion-control",
      -  "gemini-omni-video",
      -  "gemini-omni-flash-1-1",
      -  "minimax-h3",
      -  "sora-2",
      -  "grok-imagine",
      -  "pruna-avatar",
      -  "heygen-avatar-iv"
      -]New value: +[
      +  "veo-3.1",
      +  "ltx-2.3",
      +  "ltx-retake",
      +  "wan-2.7",
      +  "seedance-2",
      +  "seedance-2.5",
      +  "seedance-2-mini",
      +  "happyhorse",
      +  "kling",
      +  "kling-3",
      +  "kling-motion-control",
      +  "kling-3-motion-control",
      +  "gemini-omni-video",
      +  "gemini-omni-flash-1-1",
      +  "minimax-h3",
      +  "wan-3",
      +  "wan-3-prime",
      +  "sora-2",
      +  "grok-imagine",
      +  "pruna-avatar",
      +  "heygen-avatar-iv"
      +]
    • changedInput schema / properties / reference_audio_urls / description
      Previous value: -"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks (≤15s combined); seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside)."New value: +"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks (≤15s combined); seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside); wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, free) — name them Audio1, Audio2… in the prompt."
    • changedInput schema / properties / reference_image_urls / description
      Previous value: -"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee). veo-3.1 reference mode uses reference_images instead."New value: +"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3 / wan-3 / wan-3-prime) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee; wan-3 / wan-3-prime up to 10, free — name them Image1, Image2… in the prompt). veo-3.1 reference mode uses reference_images instead."
    • changedInput schema / properties / reference_video_urls / description
      Previous value: -"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output."New value: +"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output. wan-3 / wan-3-prime: up to 5 (each 1-15s, ≤15s combined, + output ≤ 30s), billed like minimax-h3 — name them Video1, Video2… in the prompt."
    • changedInput schema / properties / resolution / description
      Previous value: -"grok: 480p | 720p. wan: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. Others: model-specific."New value: +"grok: 480p | 720p. wan-2.7: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. wan-3 / wan-3-prime: 480p | 720p (default) | 1080p. Others: model-specific."
    • changedInput schema / properties / video_url / description
      Previous value: -"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds."New value: +"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds. wan-3 / wan-3-prime: a 1-15s clip used as reference Video1, billed as input seconds."
  5. Changed3 schema fields changed
    • changedInput schema / properties / reference_audio_urls / description
      Previous value: -"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks; seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside)."New value: +"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks (≤15s combined); seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside)."
    • changedInput schema / properties / reference_image_urls / description
      Previous value: -"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3) — character/style reference images (plural array; seedance-2.5 accepts up to 50, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee). veo-3.1 reference mode uses reference_images instead."New value: +"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3) — character/style reference images (plural array; seedance-2 / seedance-2-mini accept up to 9, seedance-2.5 up to 30, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee). veo-3.1 reference mode uses reference_images instead."
    • changedInput schema / properties / sound / description
      Previous value: -"kling / kling-3 / ltx-2.3 — generate audio alongside the video."New value: +"kling / kling-3 — generate audio alongside the video. (ltx-2.3 always renders native audio; seedance uses generate_audio.)"
  6. Changed8 schema fields changed
    • changedInput schema / properties / duration / description
      Previous value: -"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10."New value: +"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10. minimax-h3: whole seconds 4-15 (default 6)."
    • changedInput schema / properties / last_frame_url / description
      Previous value: -"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"New value: +"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1, minimax-h3 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"
    • changedInput schema / properties / model / enum
      Previous value: -[
      -  "veo-3.1",
      -  "ltx-2.3",
      -  "ltx-retake",
      -  "wan-2.7",
      -  "seedance-2",
      -  "seedance-2.5",
      -  "seedance-2-mini",
      -  "happyhorse",
      -  "kling",
      -  "kling-3",
      -  "kling-motion-control",
      -  "kling-3-motion-control",
      -  "gemini-omni-video",
      -  "gemini-omni-flash-1-1",
      -  "sora-2",
      -  "grok-imagine",
      -  "pruna-avatar",
      -  "heygen-avatar-iv"
      -]New value: +[
      +  "veo-3.1",
      +  "ltx-2.3",
      +  "ltx-retake",
      +  "wan-2.7",
      +  "seedance-2",
      +  "seedance-2.5",
      +  "seedance-2-mini",
      +  "happyhorse",
      +  "kling",
      +  "kling-3",
      +  "kling-motion-control",
      +  "kling-3-motion-control",
      +  "gemini-omni-video",
      +  "gemini-omni-flash-1-1",
      +  "minimax-h3",
      +  "sora-2",
      +  "grok-imagine",
      +  "pruna-avatar",
      +  "heygen-avatar-iv"
      +]
    • changedInput schema / properties / reference_audio_urls / description
      Previous value: -"seedance reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks; seedance-2.5: up to 10 (≤30s combined)."New value: +"reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference audio tracks; seedance-2.5: up to 10 (≤30s combined); minimax-h3: up to 3 (free; needs a reference image or video alongside)."
    • changedInput schema / properties / reference_image_urls / description
      Previous value: -"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1) — character/style reference images (plural array; seedance-2.5 accepts up to 50, the gemini-omni models up to 7). veo-3.1 reference mode uses reference_images instead."New value: +"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1 / minimax-h3) — character/style reference images (plural array; seedance-2.5 accepts up to 50, the gemini-omni models up to 7, minimax-h3 up to 9 — first 5 free, each extra adds a flat fee). veo-3.1 reference mode uses reference_images instead."
    • changedInput schema / properties / reference_video_urls / description
      Previous value: -"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation."New value: +"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation. minimax-h3: up to 3 (each 2-15s, ≤15s combined), their seconds are billed at the same per-second rate as the output."
    • changedInput schema / properties / resolution / description
      Previous value: -"grok: 480p | 720p. wan: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. Others: model-specific."New value: +"grok: 480p | 720p. wan: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. minimax-h3: 768P (default) | 2K. Others: model-specific."
    • changedInput schema / properties / video_url / description
      Previous value: -"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it)."New value: +"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it). minimax-h3: a 2-15s clip used as the motion/scene reference, billed as input seconds."
  7. Changed7 schema fields changed
    • changedInput schema / properties / duration / description
      Previous value: -"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15."New value: +"kling: \"5\" | \"10\" (string). grok: \"6\" | \"10\" (string). ltx-2.3: 6 | 8 | 10 seconds. seedance: 1-15. gemini-omni-video / gemini-omni-flash-1-1: 4 | 6 | 8 | 10."
    • changedInput schema / properties / last_frame_url / description
      Previous value: -"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"New value: +"Optional end frame for image-to-video — kling-3, veo-3.1, ltx-2.3, seedance-2, wan-2.7, gemini-omni-flash-1-1 (the models with last-frame support). Renders a start→end transition. (kling-3 also accepts an explicit image_urls array; kling 2.6 has NO end frame.)"
    • changedInput schema / properties / model / enum
      Previous value: -[
      -  "veo-3.1",
      -  "ltx-2.3",
      -  "ltx-retake",
      -  "wan-2.7",
      -  "seedance-2",
      -  "seedance-2.5",
      -  "seedance-2-mini",
      -  "happyhorse",
      -  "kling",
      -  "kling-3",
      -  "kling-motion-control",
      -  "kling-3-motion-control",
      -  "sora-2",
      -  "grok-imagine",
      -  "pruna-avatar",
      -  "heygen-avatar-iv"
      -]New value: +[
      +  "veo-3.1",
      +  "ltx-2.3",
      +  "ltx-retake",
      +  "wan-2.7",
      +  "seedance-2",
      +  "seedance-2.5",
      +  "seedance-2-mini",
      +  "happyhorse",
      +  "kling",
      +  "kling-3",
      +  "kling-motion-control",
      +  "kling-3-motion-control",
      +  "gemini-omni-video",
      +  "gemini-omni-flash-1-1",
      +  "sora-2",
      +  "grok-imagine",
      +  "pruna-avatar",
      +  "heygen-avatar-iv"
      +]
    • changedInput schema / properties / reference_image_urls / description
      Previous value: -"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse) — character/style reference images (plural array; seedance-2.5 accepts up to 50). veo-3.1 reference mode uses reference_images instead."New value: +"reference-to-video (seedance-2 / seedance-2.5 / seedance-2-mini / happyhorse / gemini-omni-video / gemini-omni-flash-1-1) — character/style reference images (plural array; seedance-2.5 accepts up to 50, the gemini-omni models up to 7). veo-3.1 reference mode uses reference_images instead."
    • changedInput schema / properties / reference_video_urls / description
      Previous value: -"seedance reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Billed on input + output seconds at the video-input rate."New value: +"seedance / gemini-omni-video reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Seedance is billed on input + output seconds at the video-input rate. gemini-omni-video: up to 3 (or pass one clip as video_url), flat price per generation."
    • changedInput schema / properties / resolution / description
      Previous value: -"grok: 480p | 720p. wan: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. Others: model-specific."New value: +"grok: 480p | 720p. wan: 1080p. seedance-2: 480p | 720p | 1080p | 4k (1080p/4k pro mode only). seedance-2.5 / seedance-2-mini: 480p | 720p. pruna-avatar: 720p (default) | 1080p. heygen-avatar-iv: 1080p only. gemini-omni-video: 720p (default) | 1080p | 4k. gemini-omni-flash-1-1: 360p | 720p (default) | 1080p | 4k. Others: model-specific."
    • changedInput schema / properties / video_url / description
      Previous value: -"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2)."New value: +"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2). gemini-omni-video / gemini-omni-flash-1-1: the clip to edit (Flash reads ≤10s of it)."
  8. Changed6 schema fields changed
    • addedInput schema / properties / draft
      Added value: +{
      +  "description": "seedance-2.5 only — draft mode: a fast, cheap 480p preview (resolution is forced to 480p). Upgrade a draft you like with type=draft-upgrade.",
      +  "type": "boolean"
      +}
    • addedInput schema / properties / draft_generation_id
      Added value: +{
      +  "description": "seedance-2.5 type=draft-upgrade only — the placeholder_id (or share link) of a finished seedance-2.5 draft. Renders that exact video at 1080p; prompt, aspect ratio and length come from the draft. Drafts can be upgraded for 7 days.",
      +  "type": "string"
      +}
    • changedInput schema / properties / mode / description
      Previous value: -"kling/kling-3: std | pro. grok: normal | fun | spicy. seedance: pro | fast. ltx-retake: replace_audio | replace_video | replace_audio_and_video. kling-motion-control / kling-3-motion-control: 720p | 1080p."New value: +"kling/kling-3: std | pro. grok: normal | fun | spicy. seedance-2: pro | fast. seedance-2.5: draft (same as draft: true). ltx-retake: replace_audio | replace_video | replace_audio_and_video. kling-motion-control / kling-3-motion-control: 720p | 1080p."
    • addedInput schema / properties / reference_video_durations
      Added value: +{
      +  "description": "Seconds per entry of reference_video_urls. Optional — the server measures the files itself for billing; supplying it only makes var2_estimate_cost exact.",
      +  "items": {
      +    "type": "number"
      +  },
      +  "type": "array"
      +}
    • changedInput schema / properties / reference_video_urls / description
      Previous value: -"seedance reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos; seedance-2.5: up to 10 (≤30s combined)."New value: +"seedance reference-to-video only — seedance-2 / seedance-2-mini: up to 3 reference videos (≤15s combined); seedance-2.5: up to 10 (≤30s combined). Billed on input + output seconds at the video-input rate."
    • changedInput schema / properties / type / enum
      Previous value: -[
      -  "text-to-video",
      -  "image-to-video",
      -  "reference-to-video",
      -  "video-to-video",
      -  "audio-to-video"
      -]New value: +[
      +  "text-to-video",
      +  "image-to-video",
      +  "reference-to-video",
      +  "video-to-video",
      +  "audio-to-video",
      +  "draft-upgrade"
      +]
  9. Changed6 schema fields changed
    • changedInput schema / properties / character_orientation / description
      Previous value: -"kling-motion-control only — take orientation from the reference image or the motion video."New value: +"kling-motion-control / kling-3-motion-control only — take orientation from the reference image or the motion video."
    • changedInput schema / properties / input_urls / description
      Previous value: -"kling-motion-control only — reference image(s) whose subject is driven by the motion video."New value: +"kling-motion-control / kling-3-motion-control only — reference image(s) whose subject is driven by the motion video."
    • changedInput schema / properties / mode / description
      Previous value: -"kling/kling-3: std | pro. grok: normal | fun | spicy. seedance: pro | fast. ltx-retake: replace_audio | replace_video | replace_audio_and_video. kling-motion-control: 720p | 1080p."New value: +"kling/kling-3: std | pro. grok: normal | fun | spicy. seedance: pro | fast. ltx-retake: replace_audio | replace_video | replace_audio_and_video. kling-motion-control / kling-3-motion-control: 720p | 1080p."
    • changedInput schema / properties / model / enum
      Previous value: -[
      -  "veo-3.1",
      -  "ltx-2.3",
      -  "ltx-retake",
      -  "wan-2.7",
      -  "seedance-2",
      -  "seedance-2.5",
      -  "seedance-2-mini",
      -  "happyhorse",
      -  "kling",
      -  "kling-3",
      -  "kling-motion-control",
      -  "sora-2",
      -  "grok-imagine",
      -  "pruna-avatar",
      -  "heygen-avatar-iv"
      -]New value: +[
      +  "veo-3.1",
      +  "ltx-2.3",
      +  "ltx-retake",
      +  "wan-2.7",
      +  "seedance-2",
      +  "seedance-2.5",
      +  "seedance-2-mini",
      +  "happyhorse",
      +  "kling",
      +  "kling-3",
      +  "kling-motion-control",
      +  "kling-3-motion-control",
      +  "sora-2",
      +  "grok-imagine",
      +  "pruna-avatar",
      +  "heygen-avatar-iv"
      +]
    • changedInput schema / properties / video_duration / description
      Previous value: -"ltx-retake: full trimmed source duration (drives per-second pricing). kling-motion-control: reference video duration (≤30s)."New value: +"ltx-retake: full trimmed source duration (drives per-second pricing). kling-motion-control: reference video duration (≤30s); kling-3-motion-control reads it server-side."
    • changedInput schema / properties / video_url / description
      Previous value: -"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control: the motion-reference video (must be pre-registered via var2)."New value: +"Source video for video-to-video. wan-2.7 / happyhorse: the clip to edit (must be uploaded to var2 first). ltx-retake: any public video URL to retake a segment of. kling-motion-control / kling-3-motion-control: the motion-reference video (must be pre-registered via var2)."
  10. Changed1 schema field changed
    • changedInput schema / properties / type / description
      Previous value: -"Pick by INPUT: an existing clip / video URL (\"restyle this clip\", \"keep the motion\") → video-to-video (video_url). ONE image to animate → image-to-video (first_frame_url; grok: reference_image_url). 2-3 reference images for a new scene (\"keep this character consistent across these references\") → reference-to-video — CALL IT with the URLs the user gave (var2 share links included) as reference_image_urls[] (veo-3.1: reference_images[]); never decline over which param name fits. Portrait + audio track (talking head, \"character talking\") → audio-to-video — when the user supplies a portrait URL and an audio URL, call immediately (pruna-avatar or heygen-avatar-iv: first_frame_url + audio_url + audio_duration_seconds); any https URL the user pasted is real and usable as-is, whatever its domain. Text only → text-to-video (default). A /video/ URL or \"clip\" means video-to-video, not image-to-video."New value: +"Pick by INPUT: an existing clip / video URL (free-form edit, \"keep the motion\") → video-to-video (video_url); swap / recast / preset restyle of a clip → var2_magic_replace instead. ONE image to animate → image-to-video (first_frame_url; grok: reference_image_url). 2-3 reference images for a new scene (\"keep this character consistent across these references\") → reference-to-video — CALL IT with the URLs the user gave (var2 share links included) as reference_image_urls[] (veo-3.1: reference_images[]); never decline over which param name fits. Portrait + audio track (talking head, \"character talking\") → audio-to-video — when the user supplies a portrait URL and an audio URL, call immediately (pruna-avatar or heygen-avatar-iv: first_frame_url + audio_url + audio_duration_seconds); any https URL the user pasted is real and usable as-is, whatever its domain. Text only → text-to-video (default). A /video/ URL or \"clip\" means video-to-video, not image-to-video."
  11. First observed

TDQS

A4.1/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations are only a title, so the description carries the full burden. It does add useful behavioral context (model defaults, per-model capability boundaries, which models accept end frames), but it never explains the job lifecycle: that creation is asynchronous, what identifier comes back, or that the agent must poll var2_get_video_result / use the webhook. For a 37-param generative job tool that silence is a real gap.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness2/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The routing rules are valuable and front-loaded, but the definition is a dense run-on stuffed with parentheticals and ALL-CAPS, and it re-lists all 23 model names that the `model` enum already enumerates verbatim — pure duplication. The kling-3/kling-3-omni/kling-4 paragraphs largely restate what those parameters' own schema descriptions say.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 37-param tool with no annotations and no output schema, the hardest agent decision — which `type` and `model` to pick — is covered exhaustively, and the schema supplies the remaining parameter detail at 97%. What is missing is the post-call story (async result retrieval, cost estimation via var2_estimate_cost), which an agent needs to actually complete the workflow.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 97%, so the baseline is 3; the description earns a 4 by adding directive semantics the schema can't (call audio-to-video immediately when a portrait URL and audio URL are supplied, "any https URL the user pasted is real and usable as-is", veo uses reference_images rather than reference_image_urls, grok uses singular reference_image_url). Some of this repeats param descriptions, but the disambiguation guidance is genuinely additive.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States the verb+resource ("Create video") and immediately scopes it by naming what it is not: replace/recast/restyle belongs to var2_magic_replace, and image animation does not belong to var2_get_image_result. An agent can distinguish this tool from every relevant sibling without opening a schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives explicit when-to-use branches keyed to input modality (existing clip -> video-to-video, one image -> image-to-video, 2-3 images -> reference-to-video, portrait+audio -> audio-to-video, text -> text-to-video) plus hard exclusions ("NEVER image-to-video for a /video/ link", "not kling_elements", "Video only — songs/music are a different tool"). It even handles the ambiguous case where a URL path contradicts the action verb.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources