opendunes
Inputs · 2 min
Copy page ▾

Video in

Send a clip with your question; get an answer about it.

  1. Pick a model that watches.
  2. Add the clip as a part of your message.
  3. Read the answer.

1. Pick a model that watches

Only some text models take video. Models that take video in lists them; the sample below uses google/gemini-2.5-flash.

Watch out

Send a clip only to a model whose list holds video. A model that does not watch cannot answer about it, and the request fails.

2. Add the clip as a part of your message

A message's content is a text, or a list of parts. To send a clip, make it a list: one text part for your question, one video_url part for the clip.

PartHolds
{"type": "text", "text": "..."}your question
{"type": "video_url", "video_url": {"url": "..."}}the clip: a public link, or the clip itself as a data:video/mp4;base64,... URL

The formats are mp4, mpeg, mov and webm.

import base64
import json
import os
import urllib.request

API_KEY = os.environ["OPENDUNES_API_KEY"]
ADDRESS = "https://opendunes.com/v1/chat/completions"
MODEL = "google/gemini-2.5-flash"
PROMPT = "Describe what happens in this clip."
CLIP_FILE = "clip.mp4"

with open(CLIP_FILE, "rb") as file:
    clip = base64.b64encode(file.read()).decode()

body = {
    "model": MODEL,
    "messages": [
        {
            "role": "user",
            "content": [
                {"type": "text", "text": PROMPT},
                {
                    "type": "video_url",
                    "video_url": {"url": "data:video/mp4;base64," + clip},
                },
            ],
        }
    ],
}

request = urllib.request.Request(
    ADDRESS,
    data=json.dumps(body).encode(),
    headers={
        "Authorization": "Bearer " + API_KEY,
        "Content-Type": "application/json",
    },
)
with urllib.request.urlopen(request) as response:
    reply = json.load(response)

print(reply["choices"][0]["message"]["content"])
print(reply["usage"]["cost"])
Do

Send a public link when the clip is online, and a data: URL when it is a file on your side. The whole request stays under 32 MB, so keep a clip you send this way short.

Don't

Don't send a link that needs a sign-in to open. The model cannot sign in, and the request fails.

3. Read the answer

The answer is the text in choices[0].message.content, as on any text request. The clip is charged as video input, counted in usage.prompt_tokens_details.video_tokens, and usage.cost is the whole charge.

To send the answer back as a stream, see Streaming. Every field of the request: Chat.

Clips made from words, not sent in, are in Video.

Good to know

A batch does not take video. Send the request on its own: see Batches.

Errors

The body says why. What to do on every status: Errors and retries.

StatusMeans
400the model takes no video, or the clip is in a format it does not read
402the balanceBalanceWhat you can spend, in DA.More on Balance and charges → cannot cover the request
404no such model
413the request is larger than the door takes