LLM node error: The video file is too long.

[libin1219]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

I uploaded a 26MB video, but the LLM node is reporting the error: "The video file is too long".

2. Additional context or comments

LLM nodes can support larger video recognition capabilities.

[dosubot[bot]]

The "video file is too long" error happens when your upload exceeds the configured video file size limit in Dify. By default, this limit is 100MB, but it can be set lower depending on your deployment or proxy settings. For your 26MB video, this error suggests the limit is set below that value or another restriction is in place.

To support larger video uploads, update these settings:

1. In your .env file (or docker-compose.yaml), set:

   UPLOAD_VIDEO_FILE_SIZE_LIMIT=100
   

   (or higher, in MB, as needed)

2. If you use Nginx or a reverse proxy, set:

   NGINX_CLIENT_MAX_BODY_SIZE=100M
   

   (or higher, matching your desired upload limit)

3. Restart your Docker containers to apply the changes:

   docker-compose down
   docker-compose up -d
   

   Reference

If you still see errors, check for other proxy or server limits, and verify the environment variables inside your running containers. Also, some models and plugins have their own input size or token limits, so very large files may still fail at the model inference step even if the upload succeeds.

For best results with very large videos, consider uploading the file to external storage and passing the URL to the model, as sending large files directly can hit model context or token limits and cause errors like "context length exceeded" or "token too long" Reference.

Supported video formats are MP4, MOV, MPEG, and MPGA. Make sure your workflow nodes are configured for multipart/form-data if using HTTP nodes for uploads Reference.

If you need to handle even larger files or run into persistent issues, using a code node to upload files externally and pass URLs is a common workaround.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[libin1219]

Your answer concerns the size of the uploaded video; my question is about processing the video size. What should I modify?

[dosubot[bot]]

Dify does not enforce a separate video processing size or duration limit in its LLM node or backend code—only the upload size limit is configurable via the UPLOAD_VIDEO_FILE_SIZE_LIMIT environment variable. Once uploaded, video files are passed directly to the model without additional checks for length or complexity in Dify itself. Any further processing limits (such as maximum video length, resolution, or token count) are imposed by the underlying model (like qwen2.5-vl-72b-instruct) or external APIs, and cannot be changed from within Dify’s configuration or codebase reference.

If you need to process larger or longer videos, make sure your upload limit is high enough, but actual processability will depend on the model’s own constraints. For very large videos, consider pre-processing (splitting, extracting frames, or uploading to external storage and passing a URL) before invoking the LLM node.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}