AGTCloud API Docs
GuidesAPI ReferenceIntegrations and resourcesUpdate and help
Video generationGemini Omni

Retrieve a Gemini Omni 1.2 Flash video

Retrieve the status and generated result of a Gemini Omni 1.2 Flash video interaction.

Use the interaction ID returned by the create API to retrieve video status, generated content and usage.

GET/v1beta/interactions/{id}

Authorization

bearerAuth
AuthorizationBearer <token>

Use an API Key from the current platform.

In: header

Path Parameters

id*string

Interaction ID returned by the create API.

Length1 <= length

Response Body

application/json

application/json

application/json

application/json

application/json

application/json

Read and save the video

HTTP 200 only means the query succeeded; check the task status. Continue polling only while the status is in_progress. When the task completes, fails or requires additional action, stop polling unchanged and handle the status accordingly. Find video results in steps entries of type model_output, then, within content, locate items with type: "video" and decode their data.

After the task succeeds, use Download video to save the file directly, or decode the Base64 data in the response yourself.

If a query encounters rate limits or temporary errors, retry with backoff.

The following Python example reads a successful response saved as response.json and decodes the video to a local file. BASE64_ENCODED_VIDEO is an illustrative marker and cannot be decoded as a video.

import base64
import json
from pathlib import Path

response = json.loads(Path("response.json").read_text(encoding="utf-8"))
video_index = 0
for step in response.get("steps", []):
    if step.get("type") != "model_output":
        continue
    for item in step.get("content", []):
        if item.get("type") == "video" and item.get("data"):
            Path(f"video-{video_index}.mp4").write_bytes(
                base64.b64decode(item["data"], validate=True)
            )
            video_index += 1

The REST response does not use the SDK convenience property output_video. To submit a new task, use Create video.

Official Google Gemini Omni get interaction documentationView the upstream protocol definition and latest field changes.