GrammarLLM
Local-first AI grammar checker running fine-tuned LLMs entirely offline.
Local Electron desktop app with real-time error correction and zero data leakage.
GrammarLLM is a desktop application that runs fine-tuned language models locally to fix grammar and assess writing quality without an internet connection. It suits students, researchers, and writers who need absolute data privacy and offline functionality.
What it is
The app runs a quantized LLM via node-llama-cpp on your machine, performing sentence-level error detection, word-level inline corrections, and quality score calculations. It generates WCAG 2.0 AA-compliant PDF reports and includes a built-in model manager supporting multiple GGUF variants.
Who it's for
It suits privacy-conscious students and researchers handling sensitive drafts who want local AI assistance without cloud APIs. It does not suit users looking for a cloud-synced web service or those without enough local disk space for model weights.
Key features
Fully Offline Operation
Runs entirely on your machine via llama.cpp with zero data sent to external servers.
Word-Level Corrections
Hover over highlighted words in your text to view and apply direct replacements, insertions, or deletions.
Multi-Backend GPU Support
Automatically detects CUDA, Metal, Vulkan, or falls back to CPU without manual configuration.
WCAG-Compliant Reports
Generate downloadable PDF assessment reports containing your writing score and all correction details.
Built-in Model Manager
Download, delete, and switch between GRMR-V3-G4B model variants or load custom GGUF URLs.
Up and running in 3 steps
Our honest verdict
What's great
- Operates completely offline, keeping your research notes and drafts entirely private.
- Provides inline word-level suggestions with automatic re-checking of edited sentences.
- Auto-detects GPU hardware across NVIDIA, Apple Silicon, and AMD/Intel chips.
- Includes a model manager to easily swap between performance-optimized or high-quality GGUF files.
Where it falls short
- Requires downloading large model files of 2 to 4 gigabytes before you can start checking text.
- Inference performance depends entirely on your local hardware specs and available GPU memory.
- Portable builds must be placed in a writable folder rather than restricted system directories.
Pricing & license
Free under its open-source license. The only cost is your own local hardware resources and disk space for the model weights.
Open remote AI-training roles
876 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
Does GrammarLLM send my text to the cloud?
No. The model runs locally on your machine via node-llama-cpp, ensuring no data ever leaves your computer.
What hardware do I need to run the app?
You need Node.js 20+ to build the app, plus enough RAM or VRAM to load a 2 to 4 GB GGUF model file.
