SuperAPI AI Directory
Partner with us中文AI newsFavorites
All toolsAI imagesdeep-research

deep-research

Deep Research

Open website

About this tool

Purpose

Deep Research is used to quickly generate in-depth research reports. It leverages multiple AI models combined with internet connectivity to generate comprehensive research analysis on various topics in about 2 minutes, accelerating the research process.

Key Capabilities

  • Fast Deep Research: Generates comprehensive research reports in about 2 minutes.
  • Multi-Platform Support: Can be quickly deployed to platforms such as Vercel and Cloudflare.
  • AI-Driven: Uses advanced AI models for accurate and insightful analysis.
  • Privacy Protection: All data is processed and stored locally in the browser, and is not uploaded to external servers (unless server-side API calls are explicitly used).
  • Multi-LLM Support: Supports Gemini, OpenAI, Anthropic, Deepseek, Atlas Cloud, Grok, Mistral, Azure OpenAI, any OpenAI-compatible LLM, OpenRouter, Ollama, and more.
  • Web Search: Supports search engines such as Searxng, Tavily, Firecrawl, fastCRW, Exa, Bocha, Brave, and more, enabling LLMs that do not support search to also use web search.
  • Thinking and Task Models: Uses "thinking" and "task" models to balance depth and speed, and supports switching research models.
  • Further Research: Allows refining or adjusting the research content at any stage of the project, and supports re-researching from that stage.
  • Local Knowledge Base: Supports uploading and processing text, Office, PDF, and other resource files to generate a local knowledge base.
  • Artifact: Supports editing research content, providing both What You See Is What You Mean (WYSIWYM) and Markdown editing modes, and allows adjusting reading level, article length, and full-text translation.
  • Knowledge Graph: Supports one-click generation of a knowledge graph to systematically understand the report content.
  • Research History: Supports saving research history so you can review it at any time and conduct further in-depth research again.
  • Local and Server API Support: Provides both local and server-side API call options.
  • SaaS and MCP Support: Can be used as a deep research service (SaaS) via SSE API, or through MCP service for other AI services.
  • PWA Support: Supports Progressive Web App (PWA), and can be used like software.
  • Multi-Key Payload: Supports Multi-Key payload to improve API response efficiency.
  • Multi-Language Support: English, Simplified Chinese, Spanish.
  • Modern Tech Stack: Built on Next.js 15 and Shadcn UI.
  • MIT License: Open source, free for personal and commercial use.

How to Use

  1. Free Use of Gemini (Recommended): Get a Gemini API Key; deploy to Vercel or Cloudflare with one click (deploying to Cloudflare requires following the relevant guide); set the LLM API key and API base URL (optional).
  2. Run Locally: Requires Node.js 18.18.0 or higher, pnpm/npm/yarn; clone the repository git clone https://github.com/u14app/deep-research.git, enter the directory cd deep-research; install dependencies pnpm install(or npm/yarn); set environment variables:cp env.tpl .env.local(development) or cp env.tpl .env(production), and fill in the variables; run the development server pnpm dev(or npm/yarn), and visit http://localhost:3000.
  3. Custom Model List: Only valid in proxy mode. Add .env or on the environment variables page NEXT_PUBLIC_MODEL_LIST, separate multiple models with commas, use the - symbol to disable a model, and use -all,+new-model-name Only specified models are allowed.
  4. Docker Deployment: Docker version must be 20 or above; note that the Docker version may lag 1-2 days behind the latest version; you can specify additional environment variables or build your own image; docker-compose.yml is supported.
  5. Static Deployment: Build a static page version and upload the out directory to a service that supports static pages (such as Github Page, Cloudflare, Vercel).
  6. Configuration: Refer to the env.tpl file to set environment variables; separate multiple keys with commas; you can set ACCESS_PASSWORD to protect the server API security; changes require redeployment to take effect.
  7. API Documentation: Supports two APIs: SSE and MCP. SSE endpoint /api/sse, recommended to use via @microsoft/fetch-event-source and listen to the message event to obtain the final report text stream. MCP supports StreamableHTTP (endpoint /api/mcp) and SSE (endpoint /api/mcp/sse) transport; a longer timeout must be set; if ACCESS_PASSWORD is set, additional request headers must be added; to enable MCP, global environment variables and the corresponding model and search engine environment variables must be set.

Applicable Scenarios

  • Quickly generate in-depth research reports on various topics.
  • Research work that requires integrating a local knowledge base and internet information.
  • Want to integrate in-depth research capabilities as a SaaS or MCP service into other AI applications.
  • Individuals or teams that value privacy and want data stored locally.

Limitations

  • No detailed information such as fees, pricing, or specific performance is provided.
  • Ollama or SearXNG may encounter TypeError: Failed to fetchdue to CORS issues, and cross-origin requests need to be configured or server proxy mode used.
  • The Docker version may lag 1-2 days behind the latest version.

Related tools

Browse category