The project formerly known as Perplexica is now Vane: a free self-hosted AI answering engine combining web search, source-linked responses and local or cloud model choices.
Vane is the current name of the project previously linked as Perplexica. The old repository address redirects to Vane, whose documentation describes an AI answering engine with web search, model configuration and source-linked responses. This listing preserves the previous name to help existing users find the same project without counting it twice.
The software is useful for people who want control over their search application and inference provider. It is still a retrieval-and-generation workflow: search coverage, source quality and the chosen model affect the result. A response containing citation markers needs the same source checking as any other AI-assisted research draft.
Research a technical option
Compare public claims
Explore an uploaded reference file
Prototype an internal search interface
Official product image. Click to inspect the details.
Install the current project
Configure models and storage
Run a bounded research question
Verify and save findings
Example prompt or task: Research this narrowly defined technical question using relevant primary documentation. Return source links, identify version assumptions and distinguish what the sources state from your interpretation.
This is an editorial research prompt. FindGoodAI has reviewed project materials, not run a controlled accuracy or speed comparison of Vane.
Local-model support provides a route without a required paid language-model API. Web searches still contact external services, and self-hosting does not make browsing offline.
Vane is MIT-licensed and can be self-hosted without an application subscription. A compatible local-model setup can avoid commercial per-request inference billing, while using your own hardware, storage and electricity.
Cloud models, hosted infrastructure and optional external services have separate terms and potential costs. The open-source project does not include unlimited paid API credits. Because online search contacts outside sources, the free local software route should not be confused with fully offline research.
The previous repository URL redirects to the current Vane project; this listing keeps both names for continuity.
The MIT software is free to self-host. Compute and connected services are separate.
The project documents local-model support, including Ollama-related setup.
No. Searching the web still requires external requests.
No. Open the cited pages and check whether they support the associated claim.
No. It is a configuration choice whose outcome depends on sources, model and task.
Review access controls first. The reviewed roadmap still lists built-in authentication as upcoming, so do not assume it is already provided.
Explore platforms, inputs and outputs, licensing, and access requirements.
The current project recommends a container route and also documents source installation. Model-server networking and persistent storage need to match the deployment environment; keep a record of the working versions and configuration.
Typical inputs are questions, conversation history and supported uploaded files. Outputs are generated answers with source metadata. API clients must use the actual provider and model identifiers returned by their instance, not illustrative identifiers copied from documentation.
Local storage and local inference can reduce external processing, but web queries, page retrieval and configured cloud models have their own data paths. Avoid including private details in queries that will be sent to public search services.
The reviewed repository lists authentication among upcoming features. Keep the instance appropriately restricted or provide reviewed access controls before remote exposure, and protect uploaded files and stored search history as working research data.
Reviewed October 3, 2026. Product facts come from the official sources below. Suggested projects, prompts and review methods are FindGoodAI editorial guidance, not measured performance results.
Loading experiences…
Saved tools are private. Approved comments are public; edits return to moderation. Comment counts include approved comments and replies. Each person has one rating. New ratings are automatically approved; withdrawn, unapproved or invalid ratings do not count. Share real experiences and avoid spam, private information or personal attacks.
For moderation appeals or data requests: support@findgoodai.com
Bring a real task and see how it fits the way you work.
User experiences