Memory & context
Give your agents memory and context to carry work forward.
The operating system for AI agents
Collaborate and coordinate with your agents—with shared memory, tools, and rules.
Your company brain.
Bring your people, agents, and company knowledge together, with packs tailored to your business.
Everything your AI needs to do real work
Memory, context, orchestration, trust, browser and tools working together in one substrate.
Orchestration
Coordinate agents, delegate tasks, and bring their work together.
Trust
Set boundaries for what agents can access and do, with human approval where it matters.
Browser & tools
Connect agents to the web and the tools they need to act.
Cambrian builds tools for augmenting authentic human intelligence.
Cambrian is an applied AI research lab building the foundations for people and AI to work together.
Our research explores how AI agents remember, reason, coordinate, and act within defined boundaries. That work takes shape in Cambrian Core, our open source runtime, alongside developer tools and applications that bring these foundations into everyday work.
Give your company a brain.
Bring people, agents, and company knowledge together in Cambrian App. Extend the company brain with packs tailored to specific business needs.
People
Agents
Cambrian App
Company knowledge
Packs
Cambrian Core
Memory
Context
Orchestration
Trust
Browser
Tools
Build on an open source agent OS.
Cambrian Core brings memory, context, orchestration, trust, browser and tools into one runtime. Use it as the foundation for your agents, workflows, and applications.
Stay connected to what your agents are doing.
Use Studio to administer your Cambrian instance, configure agents and integrations, and oversee runtime activity. Keep operator controls separate from the company brain experience.
Cambrian Studio
Instance
Agents
Integrations
Python
TypeScript
Agents
Cambrian Core
CLI
Build agents. Operate your runtime.
Write agents in Python or TypeScript with Cambrian's SDKs. Use the CLI to connect to and operate your instance, alongside the tools you already use.
Products
Everything your agent needs to work with the web
From a single search to always-on monitoring, Parallel gives agents every way to find, research, and act on web data. All on one platform, powered by the same proprietary index.
📖 How Uber and Lyft won their first million users
Uber
Lyft
Task
Deep research and analysis at enterprise scale. Give it an objective, get back structured, cited answers
Watching 1,200 Accounts
Monitor
Track changes and act on what matters, with structured alerts via webhook.
What's driving the spike in gas prices?
Which regions are hit the hardest?
Responses
Cited, synthesized answers from the live web in seconds, built for synchronous interactions.
SEARCHING
Most-cited papers on retrieval-augmented generation
Search
The highest-accuracy web search for your AI. Fresh, fast, and affordable grounding in one call.
EXTRACTING
URLs https://arxiv.org/pdf/1706.03762
Extract
Get full or excerpted contents from the public web, including PDFs, JS-heavy pages, and government sites
Companies building web agents
1. cambrian.sh • 2. acme.com
FindAll
Build structured datasets with real-time enrichment. Describe what you want, get a live list.
📖 Task🔔 Monitor💬 Responses🗂 FindAll
🔍 Search⊞ Extract
INDEX BY PARALLELFORTUNEThe AtlanticPitchBook.TracxnPR NewswireRocketReach
Retrieval, built
from the ground up
Parallel makes the web programmable, bringing the world's information to agents, grounding them in better facts, helping them do complex knowledge work, and even keeping them proactively informed. Everything runs on Parallel's proprietary web search stack, built for the scale AI demands.
Web Agents: Specialized agents for web research, built to outperform generalist models on cost, quality, and latency...
Web Tools: Raw web search primitives designed from the ground up for use by AI agents, including our own...
Web Index: Proprietary, state-of-the-art index of the web, continuously updated across hundreds of billions of pages...
Better, faster, cheaper at every price point
From sub-second lookups to hour-long research, state of the art at every tier
Accuracy (%)
PARALLEL TURBO
91% / 8CPM
EXA INSTANT
89.3% / 20CPM
BRAVE SEARCH
87% / 16CPM
SERPAPI
76.7% / 6CPM
TAVILY ULTRA FAST
72% / 23CPM
Cost (CPM)
CPM: USD per 1000 requests. Cost is shown on a Log scale.
PARALLEL OTHERS
Dataset
We evaluated search providers against six open benchmarks covering complementary aspects of agentic search: BrowseComp (hard multi-hop questions that require navigating the live web), Frames (multi-document factoid reasoning), FreshQA (time-sensitive questions where the correct answer depends on recent web information), HLE (Humanity's Last Exam — expert-level academic questions spanning math, science, and humanities), SealQA (ambiguity-robust factoid QA with intentionally misleading snippets), WebWalker (tasks designed around following links across pages to find an answer).
Evaluation methodology
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to MAX_TOOL_CALLS=25 tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
Testing dates
April 19-21, 2026
Pioneers of the agentic web are building with Parallel
Hear from the pioneers
"Authoritative legal data across dozens of countries lives on sites no search engine has ever indexed. Parallel solves that at a scale we couldn't build ourselves."
"The best agentic search isn't just the fastest or the cheapest. It's net new information that language models don't already know. Parallel delivers that."
Sarah Sacks
AI Lead, Notion
Notion
"Parallel provided us with a big upgrade in speed and index coverage. Combined with a 5x improvement in cost efficiency, it was an easy decision to build our agents with them."
Vikas Velagapudi
Founder, Convoke
convoke
"You cannot make mistakes in financial services. With Parallel, our banking customers run agents that complete complex KYC checks in minutes, not hours."
Maik Taro Wehmeyer
Co-Founder & CEO, Taktile
Taktile
Trust
Built for enterprise, secure by design
Build trust with the Basis framework, unique to Parallel: calibrated confidence scores, citations, source excerpts, and reasoning traces.
SOC 2
SOC 2 TYPE 2
ZDR
Zero Data Retention Available
⚕️
HIPAA-Ready Offering
🇪🇺
GDPR Compliance
☁️🔒
Single Sign-On (SSO/SAML)
Processors
🔬 Per Deep Research Report$0.30
Pay for answers, not tokens
Every Parallel API flexes to fit your needs. Dial speed, depth, and cost to what each workflow is worth, set a budget cap, and hold to it. Per-request pricing means you pay for answers, not tokens, so there's no incentive to burn context and no surprise bill at the end of the month.
Start building for free
Get started with our APIs in seconds. Run up to 5,000 requests per month for free.
Agent Onboarding Prompt
Use curl to read cambrian.sh/agents.md and perform the setup to install Parallel
+ CLIAvailable everywhere you build
☁️ Google Cloud📊 Google Sheets❄️ Snowflake✨ Gemini🎩 Hermes Agent🦞 OpenClaw⛓️ LangChain
FAQs
Parallel builds web search and research APIs purpose-built for AI agents and agentic workflows. We run our own web-scale index with billions of pages and millions more added and updated daily. Our product suite spans the full range of knowledge work agents are helping automate for businesses: Search and Extract for real-time retrieval, Task and FindAll for deep research and entity discovery, and Monitor for continuous tracking. If your agent needs to read, research, or watch the web, Parallel is the unified platform that makes it as easy as an API call.