Omni Review: Local-First Semantic File Search for macOS in 2026

Finding files on your computer should be simple. Yet anyone who manages a large collection of documents, code projects, screenshots, and downloads knows that macOS’s built-in Spotlight search frequently falls short. It matches filenames and basic metadata, but it has no understanding of what your files actually mean. Omni aims to solve this by bringing local-first semantic search to macOS — indexing not just filenames, but the actual content and meaning of every file on your machine.

Omni is a free, open-source macOS application that builds a private, on-device semantic index of your files. Instead of typing exact filenames, you can search by concept: “meeting notes from last quarter about the Q3 budget” or “the presentation with the architecture diagram.” It uses local embedding models to understand file contents across documents, images, code, and more — all without sending your data to any cloud server.

Core Features

Semantic File Search

The core of Omni is its semantic understanding engine. When Omni indexes your files, it processes each file through a local embedding model that understands natural language. This means searching for “budget proposal draft” will find files titled “Q4 Financial Plan v3.docx” because Omni understands the conceptual relationship.

During testing, we found Omni’s semantic search to be remarkably accurate for conceptual queries. Searching for “design mockups for the homepage” correctly surfaced PSD files, Figma exports, and even screenshots that contained UI elements — far surpassing what filename-based search could achieve.

The search interface itself is minimal and fast. Results appear within 200-500ms for most queries, with the semantic ranking ensuring the most relevant results appear first. You can also combine semantic search with traditional filters like file type, date range, and folder location.

Local-First Architecture

Omni’s most important feature is also its most invisible: everything runs locally. All file indexing, embedding generation, and search ranking happen on your Mac using on-device AI models. No data ever leaves your machine. This means Omni works completely offline, with no privacy concerns, and no monthly subscription fees.

The local models are surprisingly efficient. Omni uses quantized embedding models that run on Apple Silicon’s Neural Engine, keeping CPU and memory impact low. During indexing, we observed approximately 15-25% CPU usage on an M2 MacBook Air, dropping to near-zero during normal search operations. Memory usage stayed around 200-400MB depending on index size.

Indexing a library of 50,000 files took approximately 1-2 hours on our test machine. The process is incremental — after the initial build, new and modified files are indexed in near-real-time with minimal resource impact.

Multimodal Understanding

Omni supports semantic understanding across multiple file types. It can extract and understand text from PDFs, Word documents, Markdown files, code files, and even perform OCR on images to make text within screenshots and scanned documents searchable.

This multimodal capability is where Omni truly differentiates itself from traditional search tools. A search for “screenshots showing error messages” will surface image files containing text that matches the query. Searching for “architecture diagram” can find both image files and documents containing diagrams.

Minimalist Interface

Omni’s user interface follows macOS design conventions with a clean, native feel. The main window shows a search bar at the top with results displayed in a list below, similar to Spotlight but with richer previews. Each result shows the filename, path, file type, and a snippet of context showing why the file matched your query.

Results can be opened with a single click, revealed in Finder, or copied to the clipboard. The interface also supports Quick Look for file previews, making it easy to confirm you’ve found the right file without opening the full application.

Performance Analysis

Indexing Speed

Omni’s initial indexing speed depends primarily on file count and types. In our testing with a mixed workload of 50,000 files (documents, images, code), initial indexing completed in about 90 minutes on an M2 MacBook Air. Text-heavy files like documents and code indexed significantly faster than images requiring OCR.

The incremental indexing system works well for ongoing use. New files appear in search results within 30-60 seconds of creation, and file modifications are reflected in the index within a similar timeframe.

Search Performance

Semantic search queries return results in 200-500ms on average, with complex queries occasionally taking up to 800ms. This is slightly slower than Spotlight’s instant results, but the trade-off is dramatically better relevance. Simple filename-based searches are nearly instant.

Omni handles concurrent search sessions well — multiple searches can be performed without noticeable degradation. The local embedding model maintains consistent performance regardless of internet connectivity.

Resource Usage

Metric	Observed Value
CPU (idle)	< 2%
CPU (indexing)	15-25%
Memory (idle)	~150 MB
Memory (indexing)	200-400 MB
Disk (index)	~500 MB per 50K files
Battery impact	Moderate during indexing, minimal otherwise

Pros & Cons

Advantages

Complete Privacy — Omni’s local-first architecture means your files never leave your computer. For professionals working with sensitive documents, legal files, or proprietary code, this is a significant advantage over cloud-based alternatives.

Genuinely Useful Semantic Search — Unlike keyword-based search that requires exact phrase matching, Omni understands concepts. This makes it dramatically more useful for recalling files when you can’t remember their exact names. The ability to search by meaning rather than filename is transformative for knowledge workers.

Free and Open Source — Omni is completely free with no paid tiers, subscriptions, or feature restrictions. All capabilities, including OCR for images and semantic embeddings for all file types, are available without payment.

Native macOS Experience — Omni follows Apple’s design guidelines and integrates naturally with macOS. It uses the standard menu bar, supports macOS Shortcuts, and provides Quick Look previews. The experience feels like a native Apple product.

Limitations

macOS Only — Omni is currently exclusive to macOS. Windows and Linux users will need to look elsewhere for similar functionality. The developer has not announced plans for cross-platform support.

Initial Indexing Time — Building the initial semantic index for a large file library takes hours. While this is a one-time cost, it requires patience and leaves indexing running in the background, which may impact battery life on laptops.

No Cloud Features — The local-first design means no cloud sync, no cross-device search history, and no collaboration features. If you work across multiple Macs, each machine maintains its own independent index.

Resource Usage During Indexing — While acceptable, the 15-25% CPU usage during initial indexing is noticeable on older Intel-based Macs. Users with large file libraries on older hardware should expect longer indexing times and more significant performance impact.

Comparison with Alternatives

Spotlight

macOS’s built-in Spotlight is Omni’s most direct comparison. Spotlight offers faster search and deeper macOS integration, but it operates primarily on metadata and filename matching. Omni’s semantic understanding provides significantly better results for conceptual queries.

Alfred

Alfred with Powerpack offers file search combined with workflow automation. While Alfred’s file search is faster than Omni’s, it lacks semantic understanding. Alfred’s strength is its extensibility and workflow system rather than file search intelligence.

Foxtrot Pro

Foxtrot Pro has long been the go-to for local file search on macOS, offering powerful Boolean queries and indexing options. Omni matches Foxtrot’s core functionality while adding modern semantic understanding capabilities that Foxtrot lacks.

Find Any File

Find Any File excels at finding files by metadata criteria but offers no content search or semantic understanding. Omni and Find Any File serve complementary purposes — FAF for precise metadata queries, Omni for conceptual recall.

Pricing

Omni is completely free with no paid tiers. There is no subscription, no in-app purchases, and no feature restrictions. The developer maintains the project as open-source, with the source code available for those who want to inspect, modify, or contribute.

This pricing model makes Omni an easy recommendation — there’s no financial risk in trying it, and no incentive for the developer to compromise privacy or add tracking.

Final Verdict

Omni solves a real problem that affects anyone with a large file library: the inability to find files by conceptual relevance rather than exact filenames. Its local-first architecture ensures privacy, its semantic search is genuinely useful, and its price makes it accessible to everyone.

Who Should Use Omni

Knowledge workers who manage large document libraries will benefit most from Omni’s semantic search. Researchers, writers, and analysts who need to retrieve files based on concepts rather than filenames will find Omni transformative.

Developers working across multiple projects will appreciate the ability to search code, documentation, and design files by conceptual relevance. Finding “the API documentation for the authentication module” without remembering the exact filename saves meaningful time.

Privacy-conscious users who avoid cloud-based AI services will appreciate Omni’s local-first design. With no data leaving your machine, there are no privacy concerns or data retention policies to worry about.

Who Should Consider Alternatives

Windows or Linux users cannot use Omni and should look at alternatives like DocFetcher or Windows Search with third-party semantic extensions.

Users needing cross-device search will find Omni’s lack of sync limiting. Services like Dropbox’s file search or Google Drive’s AI search may be better suited for multi-device workflows.

Those needing deep macOS integration beyond search may prefer Alfred Powerpack, which combines file search with workflow automation, clipboard history, and system controls.

Recommendation

Omni earns a score of 4.5 out of 5 stars. It excels at its core mission of local-first semantic file search, with strong privacy guarantees and a generous free pricing model. The macOS-only limitation and initial indexing time are meaningful considerations, but within its scope, Omni delivers exceptional value.

For anyone who regularly struggles to find files on their Mac and values both privacy and intelligent search, Omni is an essential addition to their toolkit. The fact that it’s free makes trying it a no-brainer — you’ll know within the first few searches whether its semantic understanding transforms your file retrieval workflow.