Assessing the Oil Crisis with Image Context

The Strait of Hormuz Crisis, Seen Through Text, Image, and Geo Coordinates

We used AskNews' new Image Context feature to assess the IEA's historic 400-million-barrel reserve release. Here's what we found when news images and text worked together.

Multimodal Analysis

Can 400 Million Barrels Save the Oil Market? What Text and Images Together Reveal

The IEA just announced the largest coordinated reserve release in history—more than double the 2022 Ukraine response. We ran an analysis using AskNews' new Image Context feature, combining text enrichments with image metadata from satellite photos, press imagery, and naval surveillance coverage. Here's what the full picture looks like when you can see and read the news simultaneously.

The Verdict: Necessary, But the Math Doesn't Add Up

The IEA can release a maximum of 2–2.5 million barrels/day. The Strait closure removed ~15 million barrels/day of crude and refined products from global supply. That leaves a net deficit exceeding 10 million barrels/day—a gap no reserve release can close.

• Markets aren't convinced: Brent rose 4.76% to $91.98 on the day of the announcement. SPI AM's Stephen Innes called it "pouring water into a leaking barrel."
• It did prevent the worst case: Without intervention, Goldman Sachs projected $150+/barrel. The release capped prices in the $85–$92 range—painful, but not catastrophic.
• Time is limited: The 400 million barrels buy weeks, not months. Macron himself acknowledged it covers only "20 days of the volume exported through the Strait." If the blockade persists, reserves deplete by end of March—and then must be replenished, adding future pressure.
• Reopening remains remote: The U.S. Navy has declined to escort commercial vessels. A retired French Vice Admiral called sending ships into active hostilities "suicidal." As of March 11, 17 tankers had been struck, and traffic through the Strait dropped from ~100 vessels/day to fewer than 5.

Bottom line: The release buys time for diplomacy, but cannot substitute for physically reopening the Strait of Hormuz.

What Images Added to the Analysis

Text Told Us the Strait Was Closed. Images Showed Us What That Looks Like.

This analysis was our first real stress test of Image Context—AskNews' new capability that automatically extracts metadata from every news image: who's pictured, what they're doing, products and objects shown, sentiment, location, and more. These enrichments are indexed alongside text, so search and retrieval now crosses both modalities simultaneously.

Here's what image enrichments surfaced that text alone missed:

• Satellite imagery of tanker damage and naval blockade positions confirmed the severity of the closure before official reports caught up—image metadata identified vessel types and geolocations automatically
• Photos of world leaders (Macron, IEA's Fatih Birol) at emergency press conferences were tagged with names, roles, and emotional sentiment—adding urgency context that headline text sanitized away

IEA's Fatih Birol on March 6—Image Context auto-identified the speaker, role, and setting (source: AFP)

• Storage facility images from Iraq and Kuwait surfaced articles about saturation risks that text-only search ranked lower, because the written content focused on policy while the photos showed the logistics reality
• Naval vessel identification in press photos helped cross-reference IRGC force deployment patterns across dozens of sources—even when articles didn't name the specific boat classes in text

Goldman Sachs' $150/barrel warning wasn't hypothetical—satellite images showed 17 damaged tankers and near-zero commercial traffic. Gulf state storage was visibly maxing out. Diplomatic tension was visible in leaders' expressions at emergency summits. Together, text and image produced an analysis that neither modality could achieve alone.

Every one of these enrichments comes paired with geo coordinates and named entities—so you don't just know what's in the image, you know where it was captured and who or what is involved. Plot tanker incidents on a map, cluster coverage by location, or filter by entity—all from a single API response.

Geo coordinates extracted from news articles and images—every enrichment is spatially grounded (source: AskNews)

How Cross-Modal Search Works

Search for "IRGC fast attack boats Strait of Hormuz"—and AskNews doesn't just find articles mentioning them in text. It also surfaces articles where press or satellite photos show IRGC naval vessels, even if the article text never names the boat class. The image metadata—vessel type, location coordinates, military insignia—gets matched to your query and pulls the full article into your results.

Search for "oil tanker damage"—and find articles where photos show hull breaches, fire, or oil slicks, alongside the text-matched coverage. One query, both modalities, complete context.

Stop Processing HTML and Images Yourself

Until now, extracting meaning from news images meant building your own pipeline: scraping HTML, parsing <img> tags, downloading images, running them through vision models, and then somehow linking those results back to the article text. That's hours of engineering and significant compute cost—per article.

With Image Context, all of that is done for you. Every article returned by the AskNews API now includes structured image enrichments alongside the text enrichments you already rely on. No HTML parsing, no image downloading, no separate vision API calls—just one request, millisecond response times, and complete multimodal intelligence.

One API call. Text + image enrichments. Milliseconds.

Quick Start Example

from asknews_sdk import AskNewsSDK

sdk = AskNewsSDK(api_key="your_key")

response = sdk.news.search_news(
    query="IRGC fast attack boats Strait of Hormuz",
    n_articles=10,
    hours_back=48,
    method="kw",
)

# Image enrichments are included automatically
for article in response.as_dicts:
    print(article.title)
    print(article.image_description)
    print(article.image_url)
    print(article.entities)
    print(article.geo_coordinates)
    print(article.sentiment)
    print(article.bias)
    print(article.assets)

Make sure you upgrade: pip install --upgrade asknews

Ready to see what your news images contain?

Image Context is already live on the API. Upgrade your SDK and start getting multimodal enrichments with every search.

Need Help Getting Started?

Our support team is ready to help. Join our Discord server or reach out to us directly at [email protected].

AskNews - News, when quality matters.

Emergent Methods, LLC, Denver Colorado, USA

© 2026 Emergent Methods LLC All rights reserved.