Skip to content
Agents tracked: 501 Releases (7d): 777 Agent status: no reported issuesSince last limit reset: Claude 10d · Codex 8d For AI agents: Markdown Updated Nov 21, 2026

AI agents for research

Agents that run multi-step web research and return a report with sources, from deep research modes to open-source frameworks.

Works with:
Usage
  1. 42
    Butterfly Effect Pte. Ltd. · General · autonomous task agent

    General-purpose autonomous agent that plans and runs multi-step tasks in a cloud sandbox, such as research, slides, websites and data analysis, with web, desktop, mobile and browser-extension clients. Manus 2.0, announced in September 2026, added a new agent harness, dedicated cloud computers and a desktop studio.

    Web appDesktop appMobile app
    50 — — — — 112 Homebrew installs a monthGid 50 $20/mo
  2. 111
    Genspark Inc. · General · autonomous task agent

    AI workspace built around a Super Agent that plans and runs multi-step tasks in a sandbox with a real browser and filesystem, producing slides, sheets, documents, research and websites. It also offers a Chrome extension, an AI browser, a Claw personal agent that can run in the cloud or locally, and in-house models such as Gen-1.

    Web appDesktop appMobile app
    36 — — — — 4,509 app store ratingsGid 36 Free + $24.99/mo
  3. 184
    ByteDance · General · super agent harness (open source)

    Open-source 'super agent' harness from ByteDance that orchestrates subagents, memory, skills and sandboxes for long-horizon research, coding and content-creation tasks. Version 2.0 is a ground-up rewrite of the original Deep Research framework.

    Open sourceWeb appSlack, Teams, TelegramMCPLocal modelsOwn API key
    27 — — 85.6K 17d ago 85.6K GitHub starsGid 27 Free (OSS)
  4. —
    OpenAI · General · autonomous task agent

    ChatGPT's agent mode for longer multi-step tasks that end in finished deliverables such as documents, spreadsheets, presentations and sites. It runs in the cloud on web and mobile, and the ChatGPT desktop app can also use local files and apps with the user's permission.

    Web appDesktop appMobile app
    — — — — — No public usage numbers Free + $8/mo
  5. —
    Google · General · personal assistant agent

    Google's 24/7 personal AI agent in the Gemini app: it runs tasks, schedules and reusable skills in the background across Gmail, Calendar, Drive, Docs and other Google apps, and can browse the web through Chrome auto browse or a remote browser. It requires a Google AI Pro or Ultra subscription and is not offered in the EEA, UK, Switzerland or Nigeria.

    Web appDesktop appMobile app
    — — — — — No public usage numbers Free + $4.99/mo
  6. —
    MiniMax · General · autonomous task agent

    MiniMax's general-purpose agent for research, coding, documents, slides and multimodal creation, offered on the web at agent.minimax.io and as a desktop app. The desktop app supports agent teams, scheduled tasks, MCP servers, a built-in browser with computer use, and MiniMax-managed or custom models.

    Web appDesktop appCLIOwn API key
    — — — — — No public usage numbers Free + $18.33/mo yearly
  7. —
    Perplexity AI · General · autonomous task agent

    Perplexity's general-purpose agent that takes a goal in natural language and deploys subagents across several frontier models to research, build apps, websites and reports, and work in connected apps, including Slack and Microsoft 365. A Personal Computer mode works on local files on Mac and Windows.

    Web appDesktop appMobile app
    — — — — — No public usage numbers —

“#” is the position among products of the same kind (agents, frameworks, tooling) by Gid Score, our 0–100 composite of adoption, community, attention and momentum. Use the filters and sorting above, or switch to the table view. Methodology · All AI agents.

How research agents work

A research agent turns a question into a plan, runs many searches, reads the pages it finds, follows leads and writes up what it learned with citations. It saves hours on the gathering step. It does not remove the checking step: the agent can misread a source, rely on a weak one or miss what is behind a paywall.

What each agent does for this task

Only agents whose own documentation describes this use are listed. Checked Oct 6, 2026.

Agent What it does Source
ChatGPT Work ChatGPT's deep research runs multi-step web research and produces a cited report docs
Manus Wide Research runs many sub-agents in parallel on a large research task and combines the results docs
Perplexity Computer Searches the web, automates browsing, extracts data and combines findings into reports docs
Genspark Deep Research agent for multi-step web research and reports docs
MiniMax Agent Research workflow that collects sources, extracts evidence and writes a brief with citations docs
Gemini Spark Researches and browses across several sites to gather and compare information docs
DeerFlow Open-source project that began as a deep research tool (version 1); version 2 is a general agent harness docs

Getting a better report

  • Ask a narrow question and say what kind of sources you trust (official statistics, peer-reviewed papers, company filings).
  • Open the citations for every claim you will rely on.
  • Ask for disagreements, not only a summary: what sources contradict each other, and why.

More options by job: the best AI agents in every category.