Global News Archive & AI Article Briefs API

Turn global news into decision-ready intelligence in one run.

Phoenix is a global news archive API and AI article summarizer. Search clean, previously collected news by country, publisher, topic, and date, then export the original articles or add structured AI briefs with summaries, key points, entities, sentiment, business relevance, and risk scores.

Choose from a catalog of more than 1,100 news publishers across 150+ countries and regions. You do not need to find article URLs, configure a scraper, or provide an OpenRouter key. The default Articles mode does not use AI, so the first run is simple and inexpensive.

What can this global news scraper do?

• Search archived international news by country, publisher, topic, and date.
• Export clean article text and metadata for research, RAG, NLP, or media monitoring.
• Create structured Basic or Pro AI article briefs through OpenRouter.
• Retrieve ready-made daily country briefs built from multiple supporting articles.
• Score importance, business relevance, and risk on a clear 1–10 scale.
• Extract people, organizations, locations, keywords, topics, sectors, and sentiment.
• Schedule recurring runs and connect the results to webhooks, integrations, or the Apify API.
• Download datasets as JSON, CSV, Excel, XML, RSS, or HTML.

Unlike a generic URL summarizer, Phoenix combines news discovery, historical archive search, filtering, clean data, and AI analysis in the same Actor.

How to use Phoenix

1. Choose Articles, AI briefs, or Daily country briefs.
2. Select countries or publishers. You only need one of these.
3. Optionally add a topic and change the date range.
4. Keep the default result limit of 10 for your first run.
5. Click Start and open the Dataset when the run finishes.

For the quickest test, leave the defaults unchanged. Phoenix returns up to 10 recent AP News articles from the last 10 days.

Try AI article briefs

This small input finds recent renewable-energy coverage in South Africa and creates five economical, business-focused briefs:

{
  "output_mode": "ai_briefs",
  "countries": ["South Africa"],
  "search_query": "renewable energy",
  "brief_focus": "business",
  "brief_tier": "basic",
  "max_results": 5
}

Phoenix filters the archive before calling the model, so irrelevant articles do not create AI charges.

Choose the right news output

Result	Best for	AI add-on
Articles	Clean news datasets, RAG, NLP, research, and larger exports	None
AI briefs	Monitoring, market intelligence, risk scanning, and analyst queues	Basic or Pro
Daily country briefs	Fast country and topic overviews already prepared from multiple articles	None

Basic vs. Pro AI summarization

Tier	OpenRouter model	Best for	AI add-on per successful new brief
Basic	Google Gemini 2.5 Flash Lite	Routine monitoring, broad scans, and lower-cost automation	$0.006
Pro	Google Gemini 2.5 Flash	Nuanced synthesis, higher-stakes research, and analyst review	$0.012

Both tiers return the same predictable fields. Pro is the stronger choice when nuance matters more than price. Basic is the recommended starting point for recurring or high-volume monitoring.

AI mode is limited to 25 articles per run. If a complete compatible brief already exists, Phoenix can reuse it without charging the AI add-on. A failed AI generation also does not trigger the AI event charge.

Simple input options

The visual form has only nine controls. AI settings are ignored unless AI briefs is selected.

Input	What it controls	Default
output_mode	Articles, AI briefs, or daily country briefs	articles
countries	One or more countries or regions	—
websites	One or more news publishers	AP News
search_query	Optional topic or keyword for article searches	—
published_from	Beginning of the date range	10 days ago
published_to	End of the date range	Now
brief_focus	General, business, risk, policy, or technology	general
brief_tier	Basic or Pro AI quality	basic
max_results	Number of rows to return	10

Choose countries or publishers, not both. Selecting a country automatically replaces the AP News quick-start default. Daily country briefs use countries rather than publishers.

Advanced API users can also send brief_language (english or source), estimate_only, and continuation_token in JSON. These options are hidden from the visual form because most users never need them.

Global news archive input examples

Export recent articles

{
  "output_mode": "articles",
  "websites": ["AP News", "Reuters"],
  "search_query": "semiconductor investment",
  "published_from": "30 days",
  "published_to": "0 days",
  "max_results": 100
}

Create Pro risk briefs

{
  "output_mode": "ai_briefs",
  "countries": ["United Kingdom"],
  "published_from": "14 days",
  "brief_focus": "risk",
  "brief_tier": "pro",
  "max_results": 10
}

Retrieve daily country briefs

{
  "output_mode": "daily_summaries",
  "countries": ["South Africa", "Ghana"],
  "published_from": "7 days",
  "max_results": 50
}

AI brief output example

AI mode keeps the original article fields and adds a structured intelligence layer:

{
  "record_type": "article",
  "site_name": "Example News",
  "country": "South Africa",
  "article_title": "Grid investment plan advances",
  "article_body": "Normalized article text...",
  "date_published": "2026-08-01T10:00:00Z",
  "article_url": "https://example.com/grid-plan",
  "summary_status": "generated",
  "summary_brief_tier": "basic",
  "summary_title": "Grid investment plan moves forward",
  "short_summary": "Officials advanced a national electricity grid investment plan.",
  "why_it_matters": "Grid capacity affects electricity reliability and industrial growth.",
  "bullet_summary": [
    "The investment plan advanced.",
    "Electricity-grid capacity is the central focus.",
    "Implementation details remain unresolved."
  ],
  "main_topic": "energy infrastructure",
  "entities": {
    "people": [],
    "organizations": ["Energy Ministry"],
    "locations": ["South Africa"]
  },
  "sentiment": "neutral",
  "importance_score": 8,
  "business_relevance_score": 7,
  "risk_score": 5,
  "affected_sectors": ["energy", "manufacturing"],
  "summary_confidence": 0.92,
  "summary_model": "google/gemini-2.5-flash-lite"
}

The run also writes an OUTPUT summary to the default key-value store. It includes the result count, filters, pagination token, and AI generated/reused/failed counts.

Freshness and returned-result coverage

Every article row includes scraped_at (when Phoenix first collected it) and last_checked_at (the latest successful freshness check when available). This makes it easy to distinguish the article's publication date from the archive record's operational freshness.

The OUTPUT record includes archiveCoverage for article and AI-brief runs, plus summaryCoverage for daily country briefs. These fields report counts, source/country breadth, and the oldest/newest evidence timestamps for the rows returned in that run. They intentionally do not claim to measure complete archive coverage, and they do not trigger an expensive archive-wide scan.

How much does global news scraping cost?

Phoenix uses transparent pay-per-event pricing:

Event	Price
Actor start, once per run	$0.00005
Each dataset row on Free or Bronze	$0.00059
Each dataset row on Silver	$0.00049
Each dataset row on Gold or above	$0.00039
Basic AI add-on, successful new brief only	$0.006
Pro AI add-on, successful new brief only	$0.012

For example, on Free or Bronze:

• 10 articles cost about $0.00595, including the start event.
• 10 newly generated Basic briefs cost about $0.06595.
• 10 newly generated Pro briefs cost about $0.12595.

Across runs, 1,000 generated Basic briefs are about $6.39–$6.59 and 1,000 Pro briefs are about $12.39–$12.59, depending on your Apify plan, plus the small start event for each run.

Stored compatible briefs, failed AI generations, article mode, and daily country brief mode do not trigger an AI add-on. Prices shown in the Actor's Pricing tab are authoritative if they change.

For API receipts, the AI billing event names are ai-article-brief-basic and ai-article-brief-pro.

For predictable spending, begin with 5–10 results and set a maximum run cost in Apify Console or the API.

Use the global news API with Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
ACTOR_ID = "thescrapelab/Apify-The-Rise-of-the-Phoenix"

run = client.actor(
    ACTOR_ID
).call(run_input={
    "output_mode": "ai_briefs",
    "countries": ["South Africa"],
    "search_query": "renewable energy",
    "brief_focus": "business",
    "brief_tier": "basic",
    "max_results": 5,
})

items = client.dataset(run["defaultDatasetId"]).list_items().items

for item in items:
    print(item["article_title"])
    print(item.get("short_summary"))
    print(item.get("why_it_matters"))

The same Actor can run from Apify Console, the REST API, JavaScript or Python clients, schedules, webhooks, Make, Zapier, Google Sheets, and other Apify integrations.

Common use cases

• Media monitoring and daily analyst queues
• Country and regional market intelligence
• Policy, regulatory, and geopolitical scanning
• Business-impact and operational-risk detection
• PR, reputation, and competitor research
• Historical news datasets for RAG, NLP, and model evaluation
• Structured article summaries for dashboards, alerts, and automations

Limitations and responsible use

• Results are limited to publishers, countries, dates, and articles available in the hosted archive.
• This Actor searches stored news. It does not crawl arbitrary URLs during your run.
• Daily country brief mode returns existing briefs; it does not generate a new multi-article daily brief on demand.
• AI briefs can inherit omissions or errors from the source article. Scores and sentiment are model-derived signals, not professional advice.
• OpenRouter requests use Zero Data Retention-capable routing, which can temporarily reduce provider availability.
• Verify consequential claims against the linked source article before acting on them.
• Use the data in line with applicable laws, publisher terms, and your own compliance requirements.

Troubleshooting and support

• No results: widen the date range, remove the topic, or choose another country or publisher.
• Country and publisher error: keep one filter type and clear the other.
• AI result limit: lower max_results to 25 or less.
• A brief failed: check summary_error_code. Failed AI generations are not charged the AI add-on.
• More results are available: use nextContinuationToken from the OUTPUT record as continuation_token in the next API run.

For help or a reproducible data-quality issue, use the Actor page's Issues tab. Include the run ID and a redacted input example so the problem can be reproduced quickly.