KevinTex — Snip & Get, fully offline

A local, free, unlimited formula-image → LaTeX converter (a self-hosted alternative to SimpleTex). Paste, drop, or snip a screenshot of a math formula and get editable Markdown + LaTeX with a live rendered preview. Recognition runs entirely on your machine using the quantized Gemma 4 E2B-it vision-language model through llama.cpp — no cloud APIs, accounts, or quotas.

Features

• Snip button (the computer-with-+ icon) — click it, drag a screen region, and the capture is converted instantly. Fully self-contained (Pillow + tkinter region selector); no gnome-screenshot/flameshot/portal needed.
• Paste (Ctrl+V), drag-and-drop, or browse for a formula image (PNG/JPG/BMP/WEBP)
• Text sharpener with persisted Off / Auto / Strong modes — conservative contrast normalization, smart upscaling, scan denoising, and edge sharpening improve small screenshots without thresholding away formula structure.
• Draw a formula in the responsive handwriting pad, with pen/eraser, pressure-aware pointer input, stroke width, undo/redo, clear, and useful fraction/root/integral/matrix starters. Drawings go directly to local Gemma.
• Voice-to-LaTeX — dictate a formula through the microphone; Gemma turns spoken math into editable Markdown + LaTeX entirely on-device through MLX on Apple Silicon and llama.cpp on Windows/Linux.
• Image preparation tools — rotate before OCR, invert dark screenshots, and compare the original with the exact processed image sent to recognition.
• Live KaTeX preview (bundled locally — the app works with no internet at all)
• Editable output — fix the LaTeX by hand and the preview updates
• Copy formats: raw, $…$, $$…$$, \[…\], \(…\), \begin{equation}…\end{equation}, Markdown
• Auto-copy after recognition (configurable in Settings)
• Thinking mode for deeper multi-pass OCR verification on dense pages
• History with search, individual deletion, and safe clear-all controls
• Recognize again using the current source image and active Thinking setting
• Download .tex output and keyboard shortcuts (Ctrl/Cmd+K, Ctrl/Cmd+S)
• Native desktop builds for Ubuntu, Windows, and Apple Silicon macOS
• Hardware acceleration with CUDA on Linux/Windows and MLX/Metal on Apple Silicon
• Switchable backend — Gemma is the default; LOCALTEX_BACKEND=lfm-vl or LOCALTEX_BACKEND=pix2tex selects an alternative local backend. The macOS application selects LOCALTEX_BACKEND=mlx automatically.

Install

Prebuilt packages are attached to each GitHub release:

• Ubuntu: kevintex_1.2.4_all.deb
• Windows: portable KevinTex-1.2.4-windows-x64.zip containing KevinTex.exe
• Apple Silicon macOS: KevinTex-1.2.4-macOS-arm64.dmg or .zip

The Windows and macOS applications open in a native window. Model weights are not bundled; they download to the current user's application-data directory on first launch. The macOS build requires an M-series Mac and uses MLX/Metal.

Ubuntu desktop app

Two options:

System-wide (.deb):

packaging/build-deb.sh
sudo apt install ./dist/kevintex_1.2.4_all.deb

Current user only (no root):

packaging/install-user.sh

Either way you get a KevinTex entry in the applications menu with its own icon. Launching it starts the local server and opens the app in its own native window (Chrome/Chromium app mode; falls back to your browser). Closing the window stops the server. On a machine without the model environment, the first launch shows a one-time setup dialog that downloads PyTorch, llama.cpp, and the Gemma GGUF weights.

The app window's Snip button (computer-with-+ icon) opens a fullscreen region selector: drag a box around a formula and it's converted immediately — the LaTeX lands in your clipboard if auto-copy is on. No external screenshot tool required.

Run from source (dev)

./run.sh

Then open http://127.0.0.1:8321 (the script opens it for you). The first launch downloads the Gemma weights (one time); after that it is fully offline.

Manual setup (if moving to another machine)

python3 -m venv .venv
.venv/bin/pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121
.venv/bin/pip install -r requirements.txt
.venv/bin/uvicorn app:app --port 8321

Tips for best recognition

• The Gemma vision model handles full formulas, sums, integrals, matrices, and mixed text+math better than the old formula-only model. Still, crop fairly tightly around the formula for best fidelity.
• Higher-resolution, high-contrast screenshots work best.
• Leave Text sharpener on Auto for most screenshots. Strong is intended for noisy or low-contrast scans; Off preserves source pixels. Use Invert for light-on-dark captures and the Original/Processed switch to check the result.
• Use the Snip button for the fastest workflow, or bind a region-screenshot hotkey (e.g. GNOME Shift+PrtSc) to copy a region to the clipboard, then Ctrl+V into the app.

Preprocessing API

POST /api/preprocess accepts multipart image data plus preprocess (off, auto, or strong), rotation (0, 90, 180, or 270), and invert (boolean), and returns the prepared PNG without loading the OCR model. The same fields are accepted by POST /api/convert; /api/snip accepts them as query parameters. Conversion responses include the applied preprocessing metadata.

POST /api/voice accepts a 16 kHz mono WAV file in the audio multipart field and an optional thinking boolean. It is available with the default gemma backend and with LOCALTEX_BACKEND=mlx.

Performance and resource controls

KevinTex serializes access to each model context, prevents duplicate model loads, runs inference outside FastAPI's event loop, and rejects oversized uploads before image decoding. Images are closed promptly after conversion. The following optional environment variables tune local inference:

• LOCALTEX_N_CTX — context size (default 16384)
• LOCALTEX_N_GPU_LAYERS — llama.cpp GPU offload (-1 means all)
• LOCALTEX_N_BATCH — llama.cpp prompt batch size
• LOCALTEX_N_THREADS — CPU inference threads
• LOCALTEX_MAX_UPLOAD_BYTES — upload cap (default 20 MiB)
• LOCALTEX_INFERENCE_QUEUE_TIMEOUT — wait before returning busy