Google News API

The Google News API is a specialized plugin that returns Google News results (articles, topic clusters, menu navigation, related topics) as clean JSON. One HTTP call per request. Deeper browsing happens by chaining tokens (topic_token, section_token, story_token, publication_token) returned inside any response.

Key Features

• Six Driver Parameters: Search by keyword (q), browse a topic (topic_token), drill into a section (section_token), expand a story cluster (story_token), view a publication's page (publication_token), or pivot through a Knowledge Graph entity (kgmid).
• Direct Publisher URLs: news_results[].link is the publisher URL directly. No Google redirect to follow, no AMP wrapper to unwrap.
• RSS Link Resolution: Standard scrape requests can resolve news.google.com/rss/articles/... wrapper links to the real publisher article page.
• Per-Article Metadata: Bylines (source.authors), publisher icons, relative date ("3 hours ago"), ISO 8601 timestamps, and thumbnails (full + small).
• Story Clusters: Multi-outlet coverage of the same news story comes back as a cluster object with a stories[] array, same schema as flat articles.
• Sort by Relevance or Date: so=0 / so=1 controls keyword and entity searches. Other drivers use Google's editorial ordering.
• Localized Menus & Topics: menu_links[] (top nav: U.S., World, Business, Technology, …) and related_topics[] are returned localized for hl / gl.
• No Pagination Quirks: A single response returns the first page (~100 articles). Deeper browsing is done by chaining tokens, with no manual offset arithmetic.
• No Blocks or CAPTCHAs: All anti-bot measures are handled automatically by Scrape.do.

Endpoint

GET https://api.scrape.do/plugin/google/news

Request Parameters

Required

Plus exactly one driver:

Sending no driver returns 400 one of q, topic_token, section_token, story_token, publication_token, kgmid is required. Sending two returns 400 exactly one of ... may be set.

Tokens are returned inside news_results[], menu_links[], sub_menu_links[], related_topics[], and related_publications[] on every response. Chain them to navigate.

Localization

Sort (search mode only)

Example Usage

Keyword Search

curl --location --request GET 'https://api.scrape.do/plugin/google/news?token=<SDO-token>&q=openai'

import requests
import json

token = "<SDO-token>"

url = f"https://api.scrape.do/plugin/google/news?token={token}&q=openai"

response = requests.request("GET", url)

print(json.dumps(response.json(), indent=2))

const axios = require('axios');

const token = "<SDO-token>";

const url = `https://api.scrape.do/plugin/google/news?token=${token}&q=openai`;

axios.get(url)
  .then(response => {
    console.log(JSON.stringify(response.data, null, 2));
  })
  .catch(error => {
    console.error(error);
  });

package main

import (
	"fmt"
	"io/ioutil"
	"net/http"
)

func main() {
	token := "<SDO-token>"

	url := fmt.Sprintf(
		"https://api.scrape.do/plugin/google/news?token=%s&q=openai",
		token,
	)

	resp, err := http.Get(url)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	body, _ := ioutil.ReadAll(resp.Body)
	fmt.Println(string(body))
}

require 'net/http'
require 'json'

token = "<SDO-token>"

url = URI("https://api.scrape.do/plugin/google/news?token=#{token}&q=openai")

response = Net::HTTP.get(url)

puts JSON.pretty_generate(JSON.parse(response))

import java.net.HttpURLConnection;
import java.net.URL;
import java.io.BufferedReader;
import java.io.InputStreamReader;

public class GoogleNews {
    public static void main(String[] args) throws Exception {
        String token = "<SDO-token>";

        String url = String.format(
            "https://api.scrape.do/plugin/google/news?token=%s&q=openai",
            token
        );

        HttpURLConnection conn = (HttpURLConnection) new URL(url).openConnection();
        conn.setRequestMethod("GET");

        BufferedReader reader = new BufferedReader(
            new InputStreamReader(conn.getInputStream())
        );
        String line;
        StringBuilder response = new StringBuilder();
        while ((line = reader.readLine()) != null) {
            response.append(line);
        }
        reader.close();

        System.out.println(response.toString());
    }
}

using System;
using System.Net.Http;
using System.Threading.Tasks;

class Program
{
    static async Task Main()
    {
        string token = "<SDO-token>";

        string url = $"https://api.scrape.do/plugin/google/news?token={token}&q=openai";

        using HttpClient client = new HttpClient();
        string response = await client.GetStringAsync(url);

        Console.WriteLine(response);
    }
}

<?php
$token = "<SDO-token>";

$url = "https://api.scrape.do/plugin/google/news?token={$token}&q=openai";

$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);

$response = curl_exec($ch);
curl_close($ch);

echo json_encode(json_decode($response), JSON_PRETTY_PRINT);
?>

curl "https://api.scrape.do/plugin/google/news?q=openai&hl=en&gl=us&token=$TOKEN"

Localized Keyword Search

curl "https://api.scrape.do/plugin/google/news?q=bundesliga&hl=de&gl=de&token=$TOKEN"

Sort by Date

curl "https://api.scrape.do/plugin/google/news?q=openai&so=1&token=$TOKEN"

Topic Stream

curl "https://api.scrape.do/plugin/google/news?topic_token=CAAqJggKIiBDQkFTRWdvSUwyMHZNRGRqTVhZU0FtVnVHZ0pWVXlnQVAB&token=$TOKEN"

Story Cluster

curl "https://api.scrape.do/plugin/google/news?story_token=CAAqNggKIjBDQklTSGpvSmMzUnZjbmt0TXpZd1NoRUtEd2lLOWRfNUVCRllrZlFnbTlaN3F5Z0FQAQ&token=$TOKEN"

Knowledge Graph Entity

# /m/02_286 = New York City
curl "https://api.scrape.do/plugin/google/news?kgmid=/m/02_286&so=1&token=$TOKEN"

Response

Top-Level Shape

{
  "search_parameters": { ... },
  "title": "U.S.",
  "news_results": [ ... ],
  "menu_links": [ ... ],
  "sub_menu_links": [ ... ],
  "related_topics": [ ... ],
  "related_publications": [ ... ]
}

news_results is always an array (empty array when no results, never null). Other fields are present contextually:

• title: populated on topic and publication responses (e.g. "Technology", "CNN").
• menu_links: top navigation strip; same on every page (localized by hl / gl).
• sub_menu_links: sections within the current topic / publication.
• related_topics: populated on keyword searches when the query resolves to a known entity.
• related_publications: populated on publication pages.

search_parameters

{
  "engine": "google_news",
  "q": "openai",
  "google_domain": "google.com",
  "hl": "en",
  "gl": "us"
}

Optional fields (topic_token, section_token, story_token, publication_token, kgmid, so) appear when set.

news_results[]

Each entry is either a flat article or a cluster (a single story covered by multiple outlets).

Flat article

{
  "position": 2,
  "title": "OpenAI Takes Aim at Google with New Image Model",
  "link": "https://www.theinformation.com/newsletters/ai-agenda/openai-takes-aim-google-new-image-model",
  "source": {
    "name": "The Information",
    "title": "The Information",
    "icon": "https://encrypted-tbn3.gstatic.com/faviconV2?...",
    "authors": ["Stephanie Palazzolo"]
  },
  "date": "20 hours ago",
  "iso_date": "2026-04-20T14:00:00Z",
  "thumbnail": "https://tii.imgix.net/production/articles/16959/46bed976.png",
  "thumbnail_small": "https://tii.imgix.net/production/articles/16959/46bed976.png",
  "topic_token": "CAAqKAgKIiJDQkFTRXdvTkwyY3ZNVEZuWW1oeGNqaHhlaElDWlc0b0FBUAE",
  "publication_token": "CAAqLggKIihDQklTR0FnTWFoUUtFblJvWldsdVptOXliV0YwYVc5dUxtTnZiU2dBUAE"
}

Cluster

The outer entry holds only the cluster headline; individual articles live in stories[].

{
  "position": 1,
  "title": "Federal Reserve signals rate cut ahead",
  "stories": [
    {
      "position": 1,
      "title": "Federal Reserve signals rate cut ahead",
      "link": "https://www.reuters.com/markets/us/federal-reserve-signals-rate-cut-2026-04-21/",
      "source": { "name": "Reuters", "icon": "..." },
      "date": "2 hours ago",
      "iso_date": "2026-04-21T12:10:50Z",
      "thumbnail": "..."
    },
    {
      "position": 2,
      "title": "Wall Street rallies on Fed signal",
      "link": "https://www.wsj.com/...",
      "source": { "name": "The Wall Street Journal" },
      "date": "3 hours ago",
      "iso_date": "2026-04-21T11:00:00Z"
    }
  ]
}

Field reference

menu_links[]

The top navigation strip; same on every page (localized by hl / gl).

[
  { "position": 1, "title": "U.S.",      "topic_token": "CAAq..." },
  { "position": 2, "title": "World",     "topic_token": "CAAq..." },
  { "position": 3, "title": "Business",  "topic_token": "CAAq..." },
  { "position": 4, "title": "Technology","topic_token": "CAAq..." }
]

sub_menu_links[]

Present on topic and publication pages: sections within the current scope (Latest, For You, Opinion, …), each with a section_token.

related_topics[]

Populated on keyword searches when the query resolves to a known entity (person, place, organization). Each item carries a topic_token you can use to pivot to that entity's topic stream.

related_publications[]

Populated on publication pages: adjacent publishers Google suggests.

Notes

• No redirects. link is the publisher URL directly.
• Tokens are opaque. Pass them back exactly as returned. Don't parse, construct, or cache them across days.
• Tokens rotate. When a token stops working (502 unexpected response), fetch a fresh one from a recent search response and retry.
• No pagination. A single response is the first page (~100 articles for keyword searches). Deeper browsing happens by chaining tokens.

Resolving Google News RSS Article Links

Google News RSS feeds can return wrapper URLs like this:

https://news.google.com/rss/articles/<ARTICLE_ID>?oc=5

Those URLs do not redirect to the publisher in a plain HTTP fetch. Google returns a JavaScript shell page instead: status 200, no article text, and no publisher URL in the raw response body.

Pass the RSS article URL to the standard Scrape.do endpoint and Scrape.do resolves the wrapper before returning the response:

curl "https://api.scrape.do/?token=$TOKEN&url=https%3A%2F%2Fnews.google.com%2Frss%2Farticles%2FARTICLE_ID%3Foc%3D5"

The response body is the publisher's article page, not the Google wrapper page.

Important details:

• gl / hl parameters on the wrapper URL are honored when Google honors that combination.
• Resolution costs 10 credits, the same as a /plugin/google/news request.
• If the wrapper cannot be resolved, the request returns 502 and is not charged.
• A 200 response means Scrape.do returned real publisher article content, not the Google shell page.

You do not need this step for /plugin/google/news responses. The news_results[].link field is already the direct publisher URL.

Navigating with Tokens

Google News doesn't paginate. Instead, it exposes a graph of topics, sections, stories, and publications. You navigate by chaining the opaque *_token values returned in any response.

Token Types

You also have:

• q: keyword search (not a token; a string).
• kgmid: a Knowledge Graph entity ID (e.g., /m/02_286 for New York City). Stable across responses.

Workflow Patterns

Search → Topic Pivot

A keyword search response includes related_topics[] when the query resolves to an entity. Pivot into the topic to get the editorial feed for that entity instead of search relevance.

# Step 1: search
curl -s "https://api.scrape.do/plugin/google/news?q=apple&token=$TOKEN" \
  | jq '.related_topics'

# Output:
# [ { "title": "Apple Inc.", "topic_token": "CAAq..." } ]

# Step 2: pivot into the topic
TOPIC=$(curl -s "https://api.scrape.do/plugin/google/news?q=apple&token=$TOKEN" \
  | jq -r '.related_topics[0].topic_token')

curl "https://api.scrape.do/plugin/google/news?topic_token=$TOPIC&token=$TOKEN"

Topic → Section

A topic response includes sub_menu_links[] with sections like Latest, For You, Opinion, etc. Drill into one for that section's feed.

SECTION=$(curl -s "https://api.scrape.do/plugin/google/news?topic_token=$TOPIC&token=$TOKEN" \
  | jq -r '.sub_menu_links[] | select(.title=="Latest") | .section_token')

curl "https://api.scrape.do/plugin/google/news?section_token=$SECTION&token=$TOKEN"

Article → Story Cluster

Search and topic results sometimes return cluster entries: entries with a stories[] array and a story_token. Pass story_token as the driver to get the full cluster page (often more outlets than the inline stories[] snapshot).

STORY=$(curl -s "https://api.scrape.do/plugin/google/news?q=fed+rate+cut&token=$TOKEN" \
  | jq -r '.news_results[] | select(.story_token) | .story_token' \
  | head -n 1)

curl "https://api.scrape.do/plugin/google/news?story_token=$STORY&token=$TOKEN"

Article → Publication

Each article carries a publication_token. Pivot to the publisher's page to see their recent coverage.

PUB=$(curl -s "https://api.scrape.do/plugin/google/news?q=openai&token=$TOKEN" \
  | jq -r '.news_results[0].publication_token')

curl "https://api.scrape.do/plugin/google/news?publication_token=$PUB&token=$TOKEN"

Top-Level Menu

Every response includes menu_links[], Google News's top navigation strip (U.S., World, Business, Technology, Sports, Entertainment, Science, Health). Each entry exposes a topic_token. Use it to navigate without first searching.

TECH=$(curl -s "https://api.scrape.do/plugin/google/news?q=anything&token=$TOKEN" \
  | jq -r '.menu_links[] | select(.title=="Technology") | .topic_token')

curl "https://api.scrape.do/plugin/google/news?topic_token=$TECH&token=$TOKEN"

Knowledge Graph Pivot

kgmid is a stable identifier for a Knowledge Graph entity (a person, place, organization). Unlike the *_token values, it doesn't rotate, so you can store it.

# /m/02_286 = New York City, stable across responses
curl "https://api.scrape.do/plugin/google/news?kgmid=/m/02_286&so=1&token=$TOKEN"

# /m/0k8z = Apple Inc.
curl "https://api.scrape.do/plugin/google/news?kgmid=/m/0k8z&token=$TOKEN"

kgmid accepts the so (sort) parameter; use so=1 for date-sorted entity feeds.

Sort Order Across Drivers

Sending so with a non-search driver returns 400 so is only valid with q or kgmid (search mode).

Handling Stale Tokens

Tokens rotate occasionally. Symptoms:

• 502 unexpected response from the API.
• A previously-working token suddenly returns no results.

Recovery: fetch a fresh token from a recent response and retry. Don't store tokens long-term; treat them as ephemeral cursors. The two stable identifiers are q (keyword strings) and kgmid (entity IDs).

# Pseudocode for a resilient navigator
fresh_response=$(curl -s "https://api.scrape.do/plugin/google/news?q=$KEYWORD&token=$TOKEN")
fresh_topic=$(echo "$fresh_response" | jq -r '.related_topics[0].topic_token // empty')

# Use the fresh token even if you had one cached
curl "https://api.scrape.do/plugin/google/news?topic_token=$fresh_topic&token=$TOKEN"

Error Handling

{ "error": "error_code", "message": "Human readable error message" }

Common Error Codes