> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vizlook.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Get Video Contents
> The `/videos` endpoint gets the visual and audio transcriptions, video summaries, and metadata for a list of URLs.
The system will quickly return results from the cache. For uncached videos, you can configure whether to perform real-time crawling.
Sign up and receive \$10 free credit.
## OpenAPI
````yaml POST /videos
openapi: 3.1.0
info:
title: Vizlook API
description: The Vizlook OpenAPI specification
license:
name: MIT
version: 1.0.0
servers:
- url: https://api.vizlook.com
security:
- apiKeyAuth: []
paths:
/videos:
post:
description: >-
The `/search` endpoint lets you intelligently search a list of videos
that precisely ranked by the most relevant moments found within their
content.
requestBody:
content:
application/json:
schema:
$ref: '#/components/schemas/VideosRequestBody'
required: true
responses:
'200':
description: Successful response
content:
application/json:
schema:
$ref: '#/components/schemas/VideosResponse'
x-codeSamples:
- lang: bash
source: |-
curl -X POST 'https://api.vizlook.com/videos' \
-H 'x-api-key: YOUR_VIZLOOK_API_KEY' \
-H 'Content-Type: application/json' \
-d '{
"urls": [
"https://www.youtube.com/watch?v=j5vQ6oXSZpU",
"https://www.youtube.com/watch?v=123"
],
"crawlMode": "Fallback",
"contentOptions": {
"includeTranscription": true,
"includeSummary": true
}
}'
- lang: python
source: |-
# pip install vizlook
from vizlook import Vizlook
vizlook = Vizlook(api_key="YOUR_VIZLOOK_API_KEY")
response = vizlook.get_video_contents(
[
"https://www.youtube.com/watch?v=j5vQ6oXSZpU",
"https://www.youtube.com/watch?v=123",
],
crawl_mode="Fallback",
include_transcription=True,
include_summary=True,
)
print("response with original API field name: ")
print(response.to_dict())
print("response with snake case field name: ")
print(response.to_dict(use_api_field_name=False))
print("get field value with snake case key from pydantic model: ")
print(response.results)
- lang: javascript
source: |-
// npm install @vizlook/sdk
import Vizlook from "@vizlook/sdk";
const vizlook = new Vizlook({
apiKey: "YOUR_VIZLOOK_API_KEY",
});
const response = await vizlook.getVideoContents(
[
"https://www.youtube.com/watch?v=j5vQ6oXSZpU",
"https://www.youtube.com/watch?v=123",
],
{
crawlMode: "Fallback",
includeTranscription: true,
includeSummary: true,
}
);
console.log("response:", response);
components:
schemas:
VideosRequestBody:
required:
- urls
type: object
properties:
urls:
description: >-
The URLs of the page where the video is located. Currently, only
YouTube URLs are supported, such as
`https://www.youtube.com/watch?v=xxxxxx`.
type: array
items:
type: string
example:
- https://www.youtube.com/watch?v=j5vQ6oXSZpU
crawlMode:
description: >-
Video crawling strategy, defaults to `Never`.
`Never`: Get the videos from the cache, no real-time crawling.
`Fallback`: First, get the videos from the cache; if not available
in the cache, crawl the videos in real time.
`Always`: Always crawl the videos in real time.
type: string
enum:
- Never
- Fallback
- Always
example: Fallback
contentOptions:
description: Configure what video results include.
type: object
properties:
includeTranscription:
description: >-
Whether to include video transcription in the results. When not
config `contentOptions`, `includeTranscription` will be set as
true by default.
type: boolean
example: true
includeSummary:
description: Whether to include video summary in the results.
type: boolean
example: true
VideosResponse:
type: object
properties:
results:
description: Video results with fetch status.
type: array
items:
$ref: '#/components/schemas/VideoResultItem'
dollarCost:
description: Dollar cost of the request.
type: object
properties:
total:
description: Total dollar cost for your request.
type: number
example: 0.104
breakdown:
description: Breakdown of costs by operation type.
type: object
nullable: true
properties:
transcription:
description: Cost of your transcription operations.
type: number
nullable: true
example: 0.002
summary:
description: Cost of your summary operations.
type: number
nullable: true
example: 0.002
crawl:
description: Cost of your real-time crawl operations.
type: number
nullable: true
example: 0.1
example:
results:
- data:
url: https://www.youtube.com/watch?v=j5vQ6oXSZpU
title: What makes OpenAI's new GPT-5 different from past versions?
description: >-
OpenAI on Thursday launched GPT-5, the latest generation of the
artificial intelligence technology that powers ChatGPT. OpenAI
said the new model is faster and smarter than ever before.
Technology journalist Jacob Ward joins CBS News to discuss the
release.
CBS News 24/7 is the premier anchored streaming news service
from CBS News and Stations that is available free to everyone
with access to the internet and is the destination for breaking
news, live events, original reporting and storytelling, and
programs from CBS News and Stations' top anchors and
correspondents working locally, nationally and around the globe.
It is available on more than 30 platforms across mobile, desktop
and connected TVs for free, as well as CBSNews.com and
Paramount+ and live in 91 countries.
thumbnail:
url: https://i.ytimg.com/vi/j5vQ6oXSZpU/sddefault.jpg
width: 640
height: 480
author:
name: CBS News
url: https://www.youtube.com/@cbsnews
avatar: >-
https://yt3.ggpht.com/ytc/AIdro_niBFv49gSx4rr1afMZU_Pv7SeuPKO2SMHvv0Ar7OKxM4o=s88-c-k-c0x00ffffff-no-rj
publishedDate: '2025-08-07T21:07:09.000Z'
duration: 359
favicon: >-
https://www.youtube.com/s/desktop/f788da0a/img/logos/favicon_48x48.png
transcription:
videoClips:
- startTime: 0
endTime: 23
visualDescription: >-
A male presenter in a dark long-sleeved shirt and jeans
stands in a modern, brightly lit office space with plants
and wooden accents. He gestures with his hands as he
speaks. The background shows a large white screen
displaying 'GPT-5'. An inset video then appears, showing a
woman looking at a laptop, which displays a 'ChatGPT 5'
interface. The interface shows a schedule planning
assistant, detailing a day's schedule from coffee with
Grant to a flight departure, and includes 'Two quick email
actions' and a 'Light packing list'. The woman nods and
smiles. Overlapping text on screen announces 'OPENAI
RELEASES GPT-5 FIFTH GENERATION OF THE A.I. TECHNOLOGY
THAT POWERS CHATGPT'.
- startTime: 24
endTime: 43
visualDescription: >-
A male tech journalist, Jacob Ward, wearing a plaid shirt
and dark pants, stands opposite a female presenter in a
light blue suit. They are on a brightly lit studio set
with a large purple background displaying 'CBS NEWS
TECHWATCH' in white bold letters with circuit board
graphics. The female presenter holds a tablet and gestures
while speaking. A smaller screen to the left shows the
previous segment with the man presenting in the modern
office, and then a graphic that says 'Introducing
ChatGPT'.
audioClips:
- startTime: 1
endTime: 10
transcription: >-
Welcome into CBS News 24/7, OpenAI launches GPT-5, it's
the latest generation of the artificial intelligence
technology that powers ChatGPT.
speakerId: speaker_1
- startTime: 11
endTime: 23
transcription: >-
OpenAI says the new model is faster and smarter than ever
before, it comes more than two years after the release of
its predecessor, and expectations are high.
speakerId: speaker_1
summary:
overallSummary: >-
This CBS News Tech Watch segment discusses the release of
OpenAI's GPT-5, the fifth generation of the AI technology
powering ChatGPT, which is touted as faster and smarter with
enhanced reasoning capabilities. Tech journalist Jacob Ward
explains how GPT-5 aims to be a universal AI assistant, making
its own decisions and even negotiating about potentially
harmful uses. The conversation also delves into the immense
financial investment in the AI industry, with trillions of
dollars being poured into data centers and chips. Ward
expresses concerns about the increasing integration of AI into
personal lives, citing examples of using GPT for sensitive
medical information, and contrasts the companies' push for
Artificial General Intelligence with expert opinions that
current AI is still a "glorified cut and paste."
sectionSummaries:
- startTime: 0
endTime: 23
title: 'OpenAI Releases GPT-5: A New Generation of AI'
summary: >-
CBS News announces the launch of OpenAI's GPT-5, the fifth
generation of AI technology behind ChatGPT, highlighting
its increased speed and intelligence. The segment features
a demonstration of GPT-5's ability to act as a schedule
planning assistant, managing appointments, emails, and
even travel logistics. Expectations for this new model are
reportedly very high within the tech community.
- startTime: 24
endTime: 107
title: >-
Understanding GPT-5's Advanced Reasoning and
Decision-Making
summary: >-
Tech journalist Jacob Ward explains that GPT-5 is
significantly different from its predecessors, moving
beyond mere prediction to incorporate a 'reasoning model'.
This allows the AI to make its own decisions about tools
and techniques needed for tasks, effectively 'thinking'
through problems. Ward also notes GPT-5's improved ability
to avoid hallucinations and its capacity to 'negotiate'
with users regarding potentially dangerous inquiries,
demonstrating a higher level of control and ethical
consideration.
status:
url: https://www.youtube.com/watch?v=j5vQ6oXSZpU
status: Success
isLiveCrawl: false
- status:
url: https://www.youtube.com/watch?v=123
status: Fail
error:
type: CRAWL_NOT_FOUND
dollarCost:
total: 0.004
breakdown:
summary: 0.002
transcription: 0.002
VideoResultItem:
type: object
properties:
data:
description: Video result. `data` is null when `status` is `Fail`.
type: object
nullable: true
properties:
url:
description: The video page URL.
type: string
example: https://www.youtube.com/watch?v=2FMBSblpcrc
title:
description: The video title.
type: string
example: How to use one paper towel | Joe Smith | TEDxConcordiaUPortland
description:
description: The video description.
type: string
example: >-
R. P. Joe Smith served as a District Attorney in Umatilla County
and nearly won a race for Oregon Attorney General without taking
a single contribution over $99.99.
thumbnail:
description: The video thumbnail image.
type: object
properties:
url:
description: image URL
type: string
example: https://i.ytimg.com/vi/2FMBSblpcrc/sddefault.jpg
width:
description: image width
type: integer
example: 640
height:
description: image height
type: integer
example: 480
author:
description: The information of video author.
type: object
properties:
name:
description: author name
type: string
example: TEDx Talks
url:
description: author detail URL
type: string
nullable: true
example: https://www.youtube.com/@tedx
avatar:
description: author avatar URL
type: string
example: >-
https://yt3.ggpht.com/1gDPO-4YPWFErfn3yQBnv0yxpkmiud1m5CrorApkOqUYdNd9FrZQ9xB9C9VmDt4EsgArnlwqAQc=s88-c-k-c0x00ffffff-no-rj
publishedDate:
description: Video published date, ISO timestamp string
type: string
example: '2012-04-18T04:26:13.000Z'
duration:
description: Video duration in seconds
type: integer
example: 268
favicon:
description: The URL of the favicon for the search result's domain.
type: string
example: >-
https://www.youtube.com/s/desktop/f788da0a/img/logos/favicon_48x48.png
transcription:
$ref: '#/components/schemas/Transcription'
summary:
$ref: '#/components/schemas/Summary'
status:
description: Video fetch status.
type: object
properties:
url:
description: The video page URL.
type: string
example: https://www.youtube.com/watch?v=2FMBSblpcrc
status:
type: string
enum:
- Success
- Fail
example: Success
error:
description: Error message when fetch video failed.
type: object
nullable: true
properties:
type:
description: >-
Error type, check detail at [Error
Codes](/documentation/error-codes#crawl-error-type-for-%2Fvideos)
page.
type: string
enum:
- FAILED_TO_PARSE_URL
- CACHE_NOT_FOUND
- CRAWL_SOURCE_FAIL
- CRAWL_NOT_FOUND
- CRAWL_VIDEO_DURATION_EXCEEDS_LIMIT
- CRAWL_SERVER_ERROR
- CRAWL_TIMEOUT
example: CRAWL_NOT_FOUND
message:
type: string
nullable: true
isLiveCrawl:
description: Used to determine whether the video is crawled in real time.
type: boolean
nullable: true
example: true
Transcription:
description: >-
The video transcription with visual and auditory when set
`contentOptions.includeTranscription` as `true`, otherwise null.
type: object
nullable: true
properties:
videoClips:
description: Video clips with visual descriptions.
type: array
items:
$ref: '#/components/schemas/VideoClip'
audioClips:
description: Audio clips with transcriptions.
type: array
items:
$ref: '#/components/schemas/AudioClip'
Summary:
description: >-
The video summary when set `contentOptions.includeSummary` as `true`,
otherwise null.
type: object
nullable: true
properties:
overallSummary:
description: Overall summary of the video.
type: string
example: >-
Joe Smith's TEDx talk highlights the significant waste generated by
paper towel usage, with Americans alone consuming billions of pounds
annually. He proposes a simple, two-step method, 'Shake and Fold,'
to dry hands effectively using just one paper towel, significantly
reducing waste. Smith demonstrates this technique for various
dispenser types, emphasizing that proper technique can achieve
superior dryness compared to using multiple sheets. He concludes by
urging the audience to adopt this method, promising a lifelong
change in their paper towel habits and noting the substantial
positive environmental impact.
sectionSummaries:
description: Section summaries of the video.
type: array
items:
type: object
properties:
startTime:
description: Start time of the section in seconds.
type: integer
example: 0
endTime:
description: End time of the section in seconds.
type: integer
example: 7
title:
description: Title of the section.
type: string
example: Introduction to TEDx ConcordiaUPortland
summary:
description: Summary of the section.
type: string
example: >-
The video opens with the TEDx ConcordiaUPortland title card,
highlighting its independent organization and theme of
'Becoming Extraordinary.' It also credits Unconventional
Guides for sharing the content. This segment features
introductory music and brief applause, setting the stage for
the presentation.
VideoClip:
description: The video clip with visual description.
type: object
properties:
startTime:
description: Start time of the video clip in seconds.
type: integer
example: 0
endTime:
description: End time of the video clip in seconds.
type: integer
example: 7
visualDescription:
description: Visual description of the video clip.
type: string
example: >-
The video opens with a grey background featuring the red 'TEDx'
logo, followed by 'ConcordiaUPortland' in black text. Below it, 'x =
independently organized TED event' is in smaller red text. The text
'BECOMING EXTRAORDINARY' appears briefly, then a yellow-green
rectangular banner with an orange bird icon and the text
'UNCONVENTIONALGUIDES.COM' fades in. The background gradually
changes from grey to white and then to a darker shade before
transitioning to black.
AudioClip:
description: The video clip with audio transcription.
type: object
properties:
startTime:
description: Start time of the video clip in seconds.
type: integer
example: 11
endTime:
description: End time of the video clip in seconds.
type: integer
example: 26
transcription:
description: Transcription of the audio clip.
type: string
example: >-
571 million, 230,000 pounds of paper towels are used by Americans
every year.
speakerId:
description: >-
The speaker ID of the audio clip, such as 'speaker_1'. If the
speaker is identified, it will be set to the speaker's name.
type: string
example: Joe Smith
securitySchemes:
apiKeyAuth:
type: apiKey
name: x-api-key
in: header
description: API key must be provided via `x-api-key` header.
````