MetaVision API Documentation

Introduction

MetaVision is an advanced Multi-Modal AI analysis engine with a built-in content optimization pipeline and optional audio transcription. All media is automatically processed through the pipeline before VLM analysis, ensuring optimal quality and token efficiency.

MetaVision uses an asynchronous queue-based architecture. Submit, then poll for results.

Authentication

All API requests require a Bearer token:

Authorization: Bearer YOUR_API_KEY

Submit Analysis

POST /api/v1/analyze

Media Source (one required)

Optional Parameters

Submit Response (202 Accepted)

{
  "job_id": "a1b2c3d4-...",
  "status": "queued",
  "created_at": "2024-03-20T10:00:00.000Z",
  "poll_url": "/api/v1/jobs/a1b2c3d4-..."
}

Poll for Results

GET /api/v1/jobs/{job_id}

Response Statuses

Completed Response

{
  "job_id": "a1b2c3d4-...",
  "status": "completed",
  "meta": {
    "request_id": "a1b2c3d4-...",
    "model_name": "general-v2",
    "execution_time": 8.45,
    "successful_categories": 2,
    "failed_categories": 0,
    "total_tokens_in": 1500,
    "total_tokens_out": 300,
    "estimated_cost": 0.0045
  },
  "data": {
    "title": { "result": "Sunset over a mountain range" },
    "custom_category": { "result": "The mood is serene." },
    "transcript": {
      "segments": [
        { "start": 0.0, "end": 2.5, "text": "Hello world", "speaker": "SPEAKER_01" }
      ],
      "detected_language": "en",
      "language_probability": 0.997
    }
  },
  "pipeline_info": {
    "metadata": {
      "streams": [
        { "codec_type": "video", "codec_name": "h264", "width": 1280, "height": 720 },
        { "codec_type": "audio", "codec_name": "aac", "sample_rate": "44100", "channels": 2 }
      ],
      "format": { "format_name": "mov,mp4", "duration": "219.286", "size": "82989735" }
    },
    "faces": {
      "faces": [
        { "id": 1, "file": "face_001.jpg", "appearances": 3, "first_seen_time": "00:00:04" }
      ]
    },
    "download_info": {
      "title": "Video Title",
      "description": "Video description...",
      "thumbnail": "https://example.com/thumb.jpg",
      "duration": 219.0,
      "webpage_url": "https://example.com/video"
    }
  },
  "pipeline_assets": {
    "thumbnail.jpg": "https://s3.../thumbnail.jpg",
    "waveform.png": "https://s3.../waveform.png",
    "faces/face_001.jpg": "https://s3.../faces/face_001.jpg",
    "thumbnails/0001.jpg": "https://s3.../thumbnails/0001.jpg"
  }
}

Account Balance

Retrieve your usage statistics grouped by time period.

GET /api/v1/account/balance

Optional Query Parameters

Response

{
  "api_key_prefix": "e59e5c0d-...",
  "description": "Production App",
  "overall": {
    "total_requests": 1250,
    "total_success": 1200,
    "total_failed": 50,
    "total_billed": 12.50
  },
  "today": { ... },
  "week": { ... },
  "month": { ... },
  "breakdown_by_type": [
    { "media_type": "image", "request_count": 800, "success_count": 790, "failed_count": 10, "total_billed": 4.00 },
    { "media_type": "video", "request_count": 450, "success_count": 410, "failed_count": 40, "total_billed": 8.50 }
  ],
  "custom_range": null
}

When from and to are provided, the custom_range field contains the aggregated stats for that date range.

Pricing

Retrieve your configured per-category prices and available quality profiles.

GET /api/v1/account/prices

Response

{
  "price_per_image": 0.005,
  "price_per_video": 0.01,
  "price_per_audio": 0.008,
  "quality_profiles": [
    { "id": "low", "name": "Low", "billing_multiplier": 1 },
    { "id": "balanced", "name": "Balanced", "billing_multiplier": 1 },
    { "id": "high", "name": "High", "billing_multiplier": 2 },
    { "id": "ultra", "name": "Ultra", "billing_multiplier": 4 }
  ],
  "billing_note": "Each category is billed at price_per_type × quality_multiplier. Transcription adds a flat 1× price_per_type regardless of quality level."
}

Request History

List your past analysis requests with pagination and filters. You can also fetch full results or cancel queued jobs.

List Requests

GET /api/v1/account/requests

Query Parameters

Response

{
  "requests": [
    {
      "request_id": "a1b2c3d4-...",
      "status": "success",
      "media_type": "video",
      "categories": "[\"title\",\"description\"]",
      "quality_profile": "balanced",
      "billed_amount": 0.02,
      "execution_time": 8.4,
      "successful_categories": 2,
      "failed_categories": 0,
      "tokens_in": 1500,
      "tokens_out": 300,
      "agent_name": "general-v3",
      "created_at": "2024-03-20T10:00:00.000Z"
    }
  ],
  "total": 156,
  "page": 1,
  "limit": 50
}

Get Request Detail

GET /api/v1/account/requests/{request_id}

Returns the full analysis result including all category data, pipeline info, and assets. Checks active jobs first (KV), then falls back to the log database (D1) for completed historical requests.

Response (Completed)

{
  "request_id": "a1b2c3d4-...",
  "status": "completed",
  "media_type": "video",
  "agent_name": "general-v3",
  "categories": ["title", "description"],
  "quality_profile": "balanced",
  "created_at": "...",
  "meta": {
    "execution_time": 8.45,
    "tokens_in": 1500,
    "tokens_out": 300,
    "billed_amount": 0.02,
    "successful_categories": 2,
    "failed_categories": 0
  },
  "data": {
    "title": { "result": "..." },
    "description": { "result": "..." }
  },
  "pipeline_info": { ... },
  "pipeline_assets": {
    "thumbnail.jpg": "https://s3.../thumbnail.jpg",
    "waveform.png": "https://s3.../waveform.png"
  }
}

Cancel a Queued Job

POST /api/v1/account/requests/{request_id}/cancel

Cancel a job that is still in queued status. Jobs that have already started processing cannot be cancelled.

Response

{
  "success": true,
  "job_id": "a1b2c3d4-...",
  "status": "cancelled"
}

Transcription

MetaVision includes an integrated speech transcription pipeline for audio and video content. When enabled, the transcript is automatically provided to VLM agents for richer context, and included in the response.

Features

• Automatic language detection
• Word-level timestamp alignment (optional)
• Speaker diarization with speaker identification (optional)
• Configurable voice activity detection

How it works

Set transcribe: true on your request. The pipeline runs after media optimization and before VLM analysis. The transcript text is included in the VLM prompt so the model has both the media and the spoken content.

Billing

Transcription is billed as a flat 1× price_per_type per request, regardless of the quality profile multiplier. For example, if you request 3 categories with a 2× quality profile plus transcription, billing = (3 × price × 2) + (1 × price).

Parameters

Transcript Response Format

{
  "transcript": {
    "segments": [
      {
        "start": 0.0,
        "end": 2.5,
        "text": "Hello, how are you?",
        "speaker": "SPEAKER_01",
        "words": [
          {"word": "Hello,", "start": 0.1, "end": 0.5, "speaker": "SPEAKER_01"},
          {"word": "how", "start": 0.6, "end": 0.9, "speaker": "SPEAKER_01"}
        ]
      }
    ],
    "detected_language": "en",
    "language_probability": 0.997,
    "speakers": {
      "SPEAKER_01": {"name": "SPEAKER_01", "time": 2.5}
    }
  }
}

Pipeline Info & Assets

When media is processed through the optimization pipeline, the completed response includes two additional fields:

pipeline_info

A JSON object containing rich metadata extracted during optimization. The structure varies by media type and source, but commonly includes:

• metadata.streams[] — Video and audio stream details (codec, resolution, framerate, sample rate, channels, bitrate)
• metadata.format — Container format info (format name, duration, file size, overall bitrate, tags/title/description)
• faces — Face detection results including unique face IDs, appearance counts, and timestamps of first/last appearance
• download_info — Source URL metadata (title, description, thumbnail, webpage URL, duration) when media was downloaded from a web URL

pipeline_assets

A map of filename → URL for generated assets. Common assets include:

Quality Profiles

The detail parameter selects a quality profile that controls how media is optimized and how billing is calculated.

Analysis Categories

Available Models

Error Handling

Code Examples

import requests, time, base64

API_KEY = "your_api_key"
BASE = "https://metavision.vip.api.efficientstack.com/api/v1"
HEADERS = {"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"}

# URL-based submission with transcription
payload = {
    "media": "https://example.com/video.mp4",
    "media_type": "video",
    "categories": ["title", "description"],
    "detail": "balanced",
    "transcribe": True,
    "transcribe_diarize": True,
    "transcribe_align": False
}

resp = requests.post(f"{BASE}/analyze", json=payload, headers=HEADERS)
job = resp.json()
print(f"Job: {job['job_id']}")

while True:
    result = requests.get(f"{BASE}/jobs/{job['job_id']}", headers=HEADERS).json()
    print(f"Status: {result['status']}")
    if result["status"] == "completed":
        print("Results:", result["data"])
        if "transcript" in result["data"]:
            print("Transcript:", result["data"]["transcript"])
        if result.get("pipeline_info"):
            print("Pipeline Info:", result["pipeline_info"])
        if result.get("pipeline_assets"):
            print("Assets:", result["pipeline_assets"])
        break
    elif result["status"] == "failed":
        print("Error:", result.get("error"))
        break
    time.sleep(2)

# Check account balance
balance = requests.get(f"{BASE}/account/balance", headers=HEADERS).json()
print(f"Total billed: ${balance['overall']['total_billed']}")

# Check prices
prices = requests.get(f"{BASE}/account/prices", headers=HEADERS).json()
print(f"Image price: ${prices['price_per_image']}/category")

# Browse request history
history = requests.get(f"{BASE}/account/requests?limit=10", headers=HEADERS).json()
for req in history["requests"]:
    print(f"{req['request_id'][:12]}... {req['status']} {req['media_type']}")

# Get full result for a past request (includes pipeline_info and pipeline_assets)
detail = requests.get(f"{BASE}/account/requests/{job['job_id']}", headers=HEADERS).json()
print("Full result:", detail)

# Cancel a queued job
cancel = requests.post(f"{BASE}/account/requests/{job['job_id']}/cancel", headers=HEADERS).json()
print("Cancel:", cancel)

const API_KEY = 'your_api_key';
const BASE = 'https://metavision.vip.api.efficientstack.com/api/v1';
const HEADERS = {'Authorization': `Bearer ${API_KEY}`, 'Content-Type': 'application/json'};

async function analyzeMedia() {
  const resp = await fetch(`${BASE}/analyze`, {
    method: 'POST', headers: HEADERS,
    body: JSON.stringify({
      media: 'https://example.com/video.mp4',
      media_type: 'video',
      categories: ['title', 'description'],
      detail: 'balanced',
      transcribe: true,
      transcribe_diarize: true
    })
  });
  const job = await resp.json();

  while (true) {
    const result = await (await fetch(`${BASE}/jobs/${job.job_id}`, {headers: HEADERS})).json();
    if (result.status === 'completed') {
      console.log('Data:', result.data);
      console.log('Pipeline Info:', result.pipeline_info);
      console.log('Assets:', result.pipeline_assets);
      return result;
    }
    if (result.status === 'failed') throw new Error(result.error);
    await new Promise(r => setTimeout(r, 2000));
  }
}

// Account balance
async function getBalance() {
  return (await fetch(`${BASE}/account/balance`, {headers: HEADERS})).json();
}

// Prices
async function getPrices() {
  return (await fetch(`${BASE}/account/prices`, {headers: HEADERS})).json();
}

// Request history
async function getHistory(page = 1) {
  return (await fetch(`${BASE}/account/requests?page=${page}`, {headers: HEADERS})).json();
}

// Request detail (includes pipeline_info and pipeline_assets)
async function getRequestDetail(id) {
  return (await fetch(`${BASE}/account/requests/${id}`, {headers: HEADERS})).json();
}

// Cancel a queued job
async function cancelJob(id) {
  return (await fetch(`${BASE}/account/requests/${id}/cancel`, {method:'POST', headers: HEADERS})).json();
}

analyzeMedia();

# Submit analysis
JOB=$(curl -s -X POST "https://metavision.vip.api.efficientstack.com/api/v1/analyze" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"media":"https://example.com/video.mp4","media_type":"video","categories":["title"],"detail":"balanced","transcribe":true,"transcribe_diarize":true}')

JOB_ID=$(echo $JOB | jq -r '.job_id')
while true; do
  RESULT=$(curl -s "https://metavision.vip.api.efficientstack.com/api/v1/jobs/$JOB_ID" -H "Authorization: Bearer YOUR_API_KEY")
  STATUS=$(echo $RESULT | jq -r '.status')
  echo "Status: $STATUS"
  if [ "$STATUS" = "completed" ] || [ "$STATUS" = "failed" ]; then echo $RESULT | jq .; break; fi
  sleep 2
done

# View pipeline info from completed result
echo $RESULT | jq '.pipeline_info'
echo $RESULT | jq '.pipeline_assets'

# Account balance
curl -s "https://metavision.vip.api.efficientstack.com/api/v1/account/balance" -H "Authorization: Bearer YOUR_API_KEY" | jq .

# Account balance with date range
curl -s "https://metavision.vip.api.efficientstack.com/api/v1/account/balance?from=2024-01-01&to=2024-03-31" -H "Authorization: Bearer YOUR_API_KEY" | jq .

# Prices
curl -s "https://metavision.vip.api.efficientstack.com/api/v1/account/prices" -H "Authorization: Bearer YOUR_API_KEY" | jq .

# Request history
curl -s "https://metavision.vip.api.efficientstack.com/api/v1/account/requests?page=1&limit=10" -H "Authorization: Bearer YOUR_API_KEY" | jq .

# Request detail (includes pipeline_info and pipeline_assets)
curl -s "https://metavision.vip.api.efficientstack.com/api/v1/account/requests/$JOB_ID" -H "Authorization: Bearer YOUR_API_KEY" | jq .

# Cancel a queued job
curl -s -X POST "https://metavision.vip.api.efficientstack.com/api/v1/account/requests/$JOB_ID/cancel" -H "Authorization: Bearer YOUR_API_KEY" | jq .