Trammar

Fix grammar and translate text completely offline on your device using native Gemini Nano, Phi-mini, or local Gemma 4 models.

Elevate your writing and translation workflow with 100% data confidentiality. Trammar (Offline AI Translator & Grammar Fixer) is a lightweight, privacy-first browser extension designed to translate text and correct grammar entirely on your local machine. By using state-of-the-art Large Language Models (LLMs) and Small Language Models (SLMs) directly in your browser, Trammar ensures that your text never leaves your device. No cloud APIs, no external network queries, no data logging. Just fast, local and secure intelligence. 🛡️ Why Choose Trammar? 100% Confidential & Secure: Perfect for enterprise employees, researchers, legal teams, and developers handling sensitive, proprietary, or personal information. Zero API Costs: Works completely free of charge. No subscriptions, API keys, or usage caps. Offline Capability: Edit, fix and translate content even when you are disconnected from the internet. Minimal Latency: Avoid network overhead and lag by running models locally using your GPU/WASM accelerators. 🤖 Adaptive Model Routing (Local Execution Tracks) Trammar detects your active browser environment and routes text processing tasks to the most efficient on-device engine: Google Chrome (Official): Utilizes native Gemini Nano via the browser's built-in LanguageModel API. Microsoft Edge: Utilizes native Phi-mini via Edge's on-device Prompt API. Other Chromium Browsers (Brave, Vivaldi, Opera, etc.): Automatically falls back to a local quantized Gemma 4 model (onnx-community/gemma-4-E2B-it-ONNX q4) running in an offscreen document with WebGPU/WASM hardware acceleration. ✨ Key Features Intuitive Inline Menu: Simply highlight text on any webpage or input field to reveal the floating quick-access menu instantly. One-Click Corrections: Instantly trigger grammar adjustments or clean translation outputs. FIFO Quick Select (MRU Queue): Remembers your four most recently used target languages for ultra-fast access. Comprehensive Language Coverage: Supports over 100 languages, including Spanish, French, German, Hindi, Chinese, Japanese, Arabic, and more. Direct Paste & Copy: Easily copy results or inject/paste corrected text directly back into active text area, input fields, or rich text editors. Sleek Dark Theme: Featuring a modern, premium dark UI designed with smooth micro-animations and custom slim scrollbars. 🔧 Setup Instructions To unlock native on-device AI capabilities, follow these simple setup steps depending on your browser: For Google Chrome (Gemini Nano) Navigate to your browser settings: Settings > System. Find "On-device AI" (Optimization Guide) and turn the feature off and back on. This will trigger the automatic download of Gemini Nano (~2.5GB). Once the model download finishes, reload your tabs. The extension status will update to "Active (Gemini Nano Available)". For Microsoft Edge (Phi-mini) Open a new tab and go to edge://flags/#edge-llm-prompt-api-for-phi-mini Set the flag to Enabled. Relaunch Microsoft Edge to initialize the native model. For Brave, Vivaldi, Opera, and others (Gemma 4 Fallback) Open the extension popup from the toolbar. Select "Other Browsers" from the dropdown. Click "Download Local Gemma 4 Engine". The extension will pull down the quantized Gemma 4 weights and cache them directly in your browser's sandboxed storage.