Molmo2 Video Captioner: AI大規模言語モデル
Generate detailed video captions with Molmo2. Choose low, medium, or high detail for temporal scene descriptions and visual summaries. Pay per 5 seconds.
Generate detailed video captions with Molmo2. Choose low, medium, or high detail for temporal scene descriptions and visual summaries. Pay per 5 seconds.
このモデルについて
Molmo2 Video Captioner turns a public video URL into a detailed written description that tracks what happens over time, not just what appears in a single frame. It can identify subjects, objects, scene changes, camera movement, interactions, and important visual context, making the result useful as accessibility alt text, a searchable media-library description, or input for downstream creative and automation workflows. Choose low detail for a concise summary, medium for a balanced caption, or high for a more comprehensive account of temporal actions and scene structure. Clips can be up to two minutes long, and the API returns the caption asynchronously so long-running video analysis fits cleanly into production pipelines, workflow nodes, and background jobs.
コスト分析
| プロバイダー | 費用 | 備考 |
|---|---|---|
| muapiapp | $0.006 minimum; $0.010 5秒あたり | Duration-based billing with a three-second minimum and 120-second maximum. |
Duration-based billing with a three-second minimum and 120-second maximum.
** 競合サービスの料金は類似のモデル構成および利用ティアに基づいて算出された推定値です。
設定スキーマ
| パラメータ | 型 | 説明 | デフォルト |
|---|---|---|---|
| 動画 URL | string | Public video URL to caption. Supports common video formats and a maximum duration of 2 minutes. | https://interactive-examples.mdn.mozilla.net/media/cc0-videos/flower.mp4 |
| Detail Level | Enum（3個の選択肢） | Caption detail: low for a brief summary, medium for a balanced description, or high for comprehensive temporal and visual detail. | medium |
Public video URL to caption. Supports common video formats and a maximum duration of 2 minutes.
https://interactive-examples.mdn.mozilla.net/media/cc0-videos/flower.mp4Caption detail: low for a brief summary, medium for a balanced description, or high for comprehensive temporal and visual detail.
medium開発者ドキュメント
Send a POST request to /api/v1/molmo2-video-captioner with a public video_url and optional detail_level of low, medium, or high. The API returns a request_id. Poll /api/v1/predictions/{id}/result until status is completed, or provide a webhook URL for asynchronous delivery.
curl -X POST "https://api.muapi.ai/api/v1/molmo2-video-captioner"
-H "x-api-key: YOUR_API_KEY"
-H "Content-Type: application/json"
-d '{
"video_url": "https://example.com/video.mp4",
"detail_level": "medium"
}'
The completed response places the generated caption in the outputs array. Clips may be up to 120 seconds long.
FAQ
It returns a written caption describing the video's subjects, objects, actions, scene changes, camera movement, and relevant visual context.
Pass a publicly reachable video URL in video_url. Common video formats are supported, and the maximum input duration is two minutes.
Low produces a concise summary, medium provides a balanced description, and high gives a more comprehensive account of temporal actions and visual details.
The request is asynchronous. Submit the video, receive a request_id, and poll the result endpoint or supply a webhook URL to receive the completed caption.
Billing is based on video duration at $0.010 per five seconds, with a $0.006 minimum for up to three seconds and a maximum billable duration of 120 seconds.
Yes. The model is available as a text-output workflow node, so its caption can feed into downstream writing, search, classification, or publishing steps.