Today, we’re launching the Parallel Search API, the most accurate web search for AI agents, built using our proprietary web index and retrieval infrastructure. Traditional search ranks URLs for humans to click. AI search needs something different: the right tokens in their context window. Parallel’s Search API is built... from the ground up for AI agents, resulting in higher accuracy, lower costs, and lower latency for end-to-end agentic workflows.show more

Parallel’s Search API is state-of-the-art, winning on both accuracy and cost across a range of benchmarks.

For the most demanding research tasks that require deep investigation across sources, Parallel Search enables agents to achieve higher accuracy using more efficient reasoning paths than all alternatives. To see the full suite of benchmarks, read the announcement blog:

The most sophisticated AI agent builders use Parallel Search. They understand that maximizing signal and minimizing noise in an agent’s context window is the single most important factor in being able to complete a task effectively. To get started and give your agents the best context from the web, visit

Need to be able to give this to Poke @interaction for my searches.

@interaction Let’s make it happen @interaction

Looks like ass

best web context = parallel 🫡

Been an active user for 2+ months, super cool product, now powering all search in @youdistro! Also sick, video, who's it by? want to get in touch with them about our upcoming launch.

That’s a good distinction

Such a big difference when you think about building for a new user!

great product. best deep research in vertical by my estimation

tried chat completions api.. response is super slow like ~ 5 sec (on avg) for a simple phone number search of an store also documentation has `/v1/chat/completions` is api url but looks like version is not needed..

Thanks for reporting, docs issue should be fixed now

Gotta say I’m a mad sucker for some nice graphic animations

Your api better not be built on top of the likes of Google search

dope paltform guys love the ai search/ctx angle ~ let's launch/spotlight you on when it helps btw

Why not just use @tavilyai that’s doing exactly this (with better security and scale) for over 2 years now

@grok does this means ULTRA8X mode will return a better deep research output than grok and other LLM models?

Amazing logo tbh. Makes me nostalgic

Glad you like it :)

Folks trying to get api access. Unfortunately not able to proceed.

Higher accuracy, lower costs, and reduced latency - the Parallel Search API ticks all the boxes for AI-driven workflows! Can't wait to integrate this into our systems and experience the difference.

does it livecrawl websites on demand? and include extended domains in the same call? been struggling with Exa lately to do livecrawl for certain websites - our team would be happy to give this a shot

If you have a specific url to crawl in mind, would recommend our extract api which will pull page contents with the ability to turn on live crawl for that page

Great potential here 🌟

Sounds cool! Curious how it handles edge cases with context, though. 🤔

Put any context in your objective! We’ve spent a bunch of time tuning to find the best web tokens based on your semantic objective

Totally agree! Context is key, otherwise, it’s just guesswork.

"proprietary web index" this is what scares most of us.... Name a new search engine that's made it on its own in the past 15 years.

We work quickly 🫡

any trial credits for prototyping??

yes! you can run up to 20k free search requests on signup

How long dod it take for u to build this out? And is it same strategy as Exa?

About 2 years. We’re the only search provider who’s built their entire stack (index, crawl, retrieval, ranking) from the ground up for what AIs need. This means we give AIs the most relevant compressed tokens from the web. Thats why our benchmarks show we’re the highest accuracy and lowest cost in the market!

Specialized search API designed for AI agents—optimizing web retrieval for machine workflows!

any support for images and summary using the web api?

Not yet, curious to learn more, what’s your use case for those features?

performing web searches using stt. it would be much more convenient if I could have images and a summary to quickly skim the information. This way, I could dive deeper into the topic if needed.

Whoa, sounds like you've cracked the code for AI search! 🔍✨ What’s next, a search that gets my snacks too? 😂

As more startups build AI copilots and agents, APIs like this become foundational. The question now is: how quickly can developers integrate it into their existing retrieval pipelines?

Would a date filter be possible? Or in the roadmap?

Stay tuned :) We do in the interim have published date in the output which you can filter on after the fact

Useful for some stuff I’m looking at, but for certain tasks with broad queries, the date filter gives the ability to ensure there’s something relevant in the first few pages.

Makes sense, it’s on the roadmap!

That's cool

🚀

AI agents need more than just ranked URLs, true.

Shipping it on @LangbaseInc soon.

@grok what does it do

But what did you get done this week