Python SDK for the AskNews API.

Supported Python versions: 3.10–3.14.

pip install asknews

• Use ask.news.search_news() for a fast, surgical, single-search lookup when you already know the keywords or entities and do not need iterative discovery.
• Use DeepNews (ask.chat.get_deep_news() in this SDK) to attach your agent to deep, iterative research across news and other sources: finding and following leads, identifying unknown entities, and connecting evidence across sources.

from asknews_sdk import AskNewsSDK

ask = AskNewsSDK(
    api_key="<YOUR API KEY>",
    scopes=["news", "chat", "stories", "analytics"]
)

query = "NVIDIA earnings"

# Fast lookup: prompt-optimized news context for your LLM.
news_context = ask.news.search_news(query).as_string

# Iterative discovery across news and other sources.
research = ask.chat.get_deep_news(
    messages=[{"role": "user", "content": "Find and follow leads on AI chip supply risks."}],
    sources=["asknews", "google", "wiki"],
    engine="v1.5",
)
print(research.choices[0].message.content)

The API doesn't stop there, checkout the full set of recipes in the Cookbook showing off a wide range of endpoints:

• /deepnews, a deep research agent that can explore the new knowledge graph, X, Reddit, Google, Wikipedia and more to build forecasts, reports, analytics, and anything else your system may need.
• /websearch, search the web and get back an LLM distillation of all the relevant web pages
• /wikipedia, high fidelity search on the Wikipedia corpus (open-sourced, https://huggingface.co/datasets/EmergentMethods/en_qdrant_wikipedia)
• /stories, high level event tracking and state of the art article clustering

Find full docs at the AskNews API documentation.

Join our Discord to see what other people are building, and to get support with your projects.