research / agent

Vane

A self-hosted AI answering engine that combines SearxNG web search with local or cloud language models.

Repository owner ItzCrazyKns

Not rated yet
0 comments

Like 0 Save

Published 11 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: A self-hosted AI answering engine that combines SearxNG web search with local or cloud language models.
Runs in: Web service
Operating systems: macOS, Linux, Windows
Needs: Runtime: Docker; Needs internet: Yes; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Yes
Tool type: Agent
Delivery: From GitHub
Version: Latest on the master branch
Access: Code on GitHub
Compatibility: According to the README, Vane runs in Docker or, without Docker, through npm, and it is built on Next.js. The README lists Ollama and cloud providers including OpenAI, Anthropic Claude, Google Gemini and Groq, and gives connection instructions for Lemonade. The README gives connection addresses for Ollama and Lemonade on Windows, Mac and Linux, and says Linux users may need to expose Ollama to the network.
External costs: The README names OpenAI, Anthropic Claude, Google Gemini and Groq as cloud providers you can connect, and Ollama and Lemonade for local models. It says you set API keys in the setup screen. It states no prices. The repository is MIT-licensed. The model provider you connect may charge for usage.
Permissions: The README's Docker command downloads and starts a background container, publishes port 3000 and creates persistent storage for data and uploaded files. The README's non-Docker route installs npm dependencies and requires SearxNG configuration. The README says users enter provider API keys and models in the setup screen. The README describes web searches, file uploads, local search history and an API for other applications. The README's Linux Ollama troubleshooting edits /etc/systemd/system/ollama.service, reloads systemd and restarts Ollama to expose it to the network. The README also describes changing browser search-engine settings, provides one-click cloud deployment links for Sealos, RepoCloud, ClawCloud and Hostinger, and says authentication is an upcoming feature.
Usage terms: MIT License, as stated in the repository. See its LICENSE file for the exact terms.
Interface: web

Code on GitHub

ItzCrazyKns/Vane

MIT

TypeScript37,283 stars4,149 forksLast pushed 1 Sept 2026 UTC

Vane is an AI-powered answering engine.

Topics: ai-agents ai-search-engine answering-engine artificial-intelligence llm machine-learning open-source-ai-search-engine perplexica

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

View on GitHub ↗

What it does

The README describes Vane as a privacy-focused AI answering engine for the user's own hardware. The README says it combines web search with local models through Ollama and cloud providers such as OpenAI, Claude and Groq, and returns answers with cited sources. The README names SearxNG as its web-search engine.

The README lists Speed, Balanced and Quality search modes, web, discussion and academic sources, weather and calculation widgets, image and video search, file uploads, domain-limited search, suggestions, a Discover feed and locally saved search history.

The README describes browser search-engine setup and a search API. The README lists additional widgets, integrations, sources, custom agents and authentication as upcoming features.

Who it is for

People who want to run their own AI answering engine on their own hardware, and developers who want to call its search API.

Example tasks

Choose Speed, Balanced or Quality mode and ask a question.
Search the web, discussions or academic papers.
Upload a document and ask questions about it.
Limit a search to specific websites.
Call the search API from your own application.

Getting started

The README recommends Docker. It says to run docker run -d -p 3000:3000 -v vane-data:/home/vane/data --name vane itzcrazykns1337/vane:latest, open the app on port 3000 of the machine and set up API keys and models in the setup screen. A slim image is available if you already run SearxNG. The README also describes building the image from source and a non-Docker route that uses npm i, npm run build and npm run start. The README has the exact commands.

Limitations

The README recommends Docker and says its image bundles SearxNG. The README says non-Docker installation and the slim image require a SearxNG instance with JSON format and Wolfram Alpha enabled. The README gives Ollama and Lemonade connection instructions for Windows, Mac and Linux. The README's Linux Ollama troubleshooting changes /etc/systemd/system/ollama.service, reloads systemd and restarts Ollama to expose it to the network. The README says local OpenAI-compatible servers must listen on 0.0.0.0 at the configured port, use the correct model name and have a nonempty API-key field even when no key is defined. The README also says Lemonade must accept connections from all interfaces. The README lists authentication as upcoming and says Tavily and Exa support is coming soon; the README also identifies Exa as a sponsor. The repository can change at any time; check its README before you install.

Data practices

The README says Vane runs on the user's hardware and saves search history locally. The README says its Docker commands create persistent volumes for data and uploaded files. Data destinations depend on the local or cloud model and search services you configure. The README names SearxNG as the web-search engine and says it accesses multiple search engines. The README does not describe telemetry or analytics.