Connect your AI to valuable web data

Improve RAG performance with the most powerful web search API

Try the API for free
Cursor logoJennie logoDatabricks logoNvidia logoImbue logollamaindex logoDDB logo
Cursor logoJennie logoDatabricks logoNvidia logoImbue logollamaindex logoDDB logo
Why Exa?

Better data, better LLM results


Exa is the first meaning-based web search powered by embeddings. It unlocks data no other search can, making your AI more relevant, factual, and reducing hallucinations.

Exa is the first meaning-based web search powered by embeddings. It unlocks data no other search can, making your AI more relevant, factual, and reducing hallucinations.

Highly customizable search


You have total control. Powerful filters such as date, category and domain. Choose # of results (1-1000s) and extractions. You name it.

You have total control. Powerful filters such as date, category and domain. Choose # of results (1-1000s) and extractions. You name it.

Get content from any webpage


Choose if you want to get just links, full parsed text content, key highlights or customizable summaries per URL.

Choose if you want to get just links, full parsed text content, key highlights or customizable summaries per URL.

Built for production and scale


High rate limits, low latency and high reliability. Ready to scale with your application.

High rate limits, low latency and high reliability. Ready to scale with your application.

Just a few lines of code

pip install exa_pyfrom exa_py import Exa exa = Exa("EXA_API_KEY") results = exa.search( "latest developments in ML", category="papers" )
{ "results": [ { "score": 0.1836702525615692, "title": "Asymptotically Optimal Regret for Black-Box Predict-then-Optimize", "id": "https://arxiv.org/abs/2406.07866", "url": "https://arxiv.org/abs/2406.07866", "publishedDate": "2024-06-12", "author": "Tan; Samuel; Frazier; Peter I" }, { "score": 0.18168984353542328, "title": "Machine Unlearning for Uplink Interference Cancellation", "id": "https://arxiv.org/abs/2406.05945", "url": "https://arxiv.org/abs/2406.05945", "publishedDate": "2024-06-10", "author": "Guven; Eray; Kurt; Gunes Karabulut" }, { "score": 0.17979080975055695, "title": "Attention as an RNN", "id": "https://arxiv.org/abs/2405.13956", "url": "https://arxiv.org/abs/2405.13956", "publishedDate": "2024-05-22", "author": "Feng; Leo; Tung; Frederick; Hajimirsadeghi; Hossein; Ahmed; Mohamed Osama; Bengio; Yoshua; Mori; Greg" }, { "score": 0.17959752678871155, "title": "FaithFill: Faithful Inpainting for Object Completion Using a Single Reference Image", "id": "https://arxiv.org/abs/2406.07865", "url": "https://arxiv.org/abs/2406.07865", "publishedDate": "2024-06-12", "author": "Mallick; Rupayan; Abdalla; Amr; Bargal; Sarah Adel" }, { "score": 0.17918667197227478, "title": "Talking Heads: Understanding Inter-layer Communication in Transformer Language Models", "id": "https://arxiv.org/abs/2406.09519", "url": "https://arxiv.org/abs/2406.09519", "publishedDate": "2024-06-13", "author": "Merullo; Jack; Eickhoff; Carsten; Pavlick; Ellie" }, { "score": 0.17634162306785583, "title": "FOOD: Facial Authentication and Out-of-Distribution Detection with Short-Range FMCW Radar", "id": "https://arxiv.org/abs/2406.04546", "url": "https://arxiv.org/abs/2406.04546", "publishedDate": "2024-06-06", "author": "Kahya; Sabri Mustafa; Sivrikaya; Boran Hamdi; Yavuz; Muhammet Sami; Steinbach; Eckehard" }, { "score": 0.17634151875972748, "title": "xLSTM: Extended Long Short-Term Memory", "id": "https://arxiv.org/abs/2405.04517", "url": "https://arxiv.org/abs/2405.04517", "publishedDate": "2024-05-07", "author": "Beck; Maximilian; Pöppel; Korbinian; Spanring; Markus; Auer; Andreas; Prudnikova; Oleksandra; Kopp; Michael; Klambauer; Günter; Brandstetter; Johannes; Hochreiter; Sepp" }, { "score": 0.17621161043643951, "title": "World Models with Hints of Large Language Models for Goal Achieving", "id": "https://arxiv.org/abs/2406.07381", "url": "https://arxiv.org/abs/2406.07381", "publishedDate": "2024-06-11", "author": "Liu; Zeyuan; Huan; Ziyu; Wang; Xiyao; Lyu; Jiafei; Tao; Jian; Li; Xiu; Furong; Xu; Huazhe" }, { "score": 0.17613205313682556, "title": "Deep learning classifier of locally advanced rectal cancer treatment response from endoscopy images", "id": "https://arxiv.org/abs/2405.03762", "url": "https://arxiv.org/abs/2405.03762", "publishedDate": "2024-05-06", "author": "Gomez; Jorge Tapias; Rangnekar; Aneesh; Williams; Hannah; Thompson; Garcia-Aguilar; Julio; Smith; Joshua Jesse; Veeraraghavan; Harini" }, { "score": 0.17593073844909668, "title": "MalleTrain: Deep Neural Network Training on Unfillable Supercomputer Nodes", "id": "https://arxiv.org/abs/2404.15668", "url": "https://arxiv.org/abs/2404.15668", "publishedDate": "2024-04-24", "author": "Ma; Xiaolong; Yan; Feng; Lei; Foster; Ian; Papka; Michael E; Liu; Zhengchun; Kettimuthu; Rajkumar" } ], "requestId": "abb0f98756d6c722ca35e06e0970652b" }

just a few lines of code, or integrate with:

LangChain logoLlamaIndex logocrewAI logo

How Customers are using Exa...

Exa's customizable search powers all sorts of customer applications. Click to experience some demos!

What you can do with Exa


Semantic Search
Keyword Search
Similarity Search
Extensive Filters
Scrape Content
1000s of Results
Long Queries
AI Summaries
Live Crawl
Recursive Crawl