Gemma
How vibe coders and indie builders use Gemma: 52 projects built with it since May 2026.
Projects built with it
52
Share of stated stacks, 4 wks
0.1%
Last 4 weeks vs 8 before
too few to tell
Add-ons built for it
0
Share of stated stacks, per week
Share rather than counts, so changes in how many posts a subreddit lets through don't show up as trends.
Often used with
- OpenAI API 14
- Ollama 13
- Qwen 13
- Gemini API 10
- Claude API 10
- Python 7
- OpenRouter 7
- Llama 7
What's built with it
Latest projects built with Gemma
- OVERRIDE — Runs a top-down shooter in the browser where an LLM villain watches the player's recent actions and taunts them while modifying the game in real time.“I vibe coded a top-down shooter where the villain is an LLM that watches how you play — and trash-talks you about it”
- TUFF — Runs large mixture-of-experts language models locally on a Mac by keeping shared weights in RAM and streaming experts from SSD through a bounded cache.“Shipped TUFF 3.0 — a Mac app that lets you run LLMs too big for your RAM”
- Warren — Captures pages you read in Chrome and builds a local, searchable markdown wiki of them using a local vision model.“Warren: self hosted browsing memory. Local LLM reads what you read, you get a searchable wiki out of it.”
- FullPrice.LOL — Searches nearby retailers and auctions for overstock, clearance, and customer return items and sends alerts when a saved search matches.“I hate paying retail so I built a search engine for overstock and customer returns”
- Kallilex — Corrects, shortens, or rephrases selected text via a global hotkey and inserts the result back in place using any configurable AI endpoint.“(Open Source): Kallilex - Correct text directly, without opening a tab!”
- YouTube Shorts content factory — Generates and publishes YouTube Shorts videos from text cards with AI-written metadata, voiceover, subtitles and music.“Here's a Youtube Shorts content factory; details on how it was made below”
- Manurio — Guides an LLM through a staged process to write a complete book, from premise and plot to chapters, revisions and export, with optional autonomous runs.“I built a local-first AI writing studio after manually testing how to write a complete book with an LLM”
- Musiclyse — Analyzes a song's audio with multiple models and lets a local LLM discuss and compare tracks with the user in a terminal chat.“I'm developing a music to LLM chatbot terminal in Python using Qwen 3.8 27B”
- Local Coding Agent — Runs as an MCP server that delegates small coding tasks from cloud coding assistants to a local Ollama model and rolls back changes that fail.“Built a local MCP bridge to stop burning subscription limits on small edits”
- Skim Recap — Watches scroll speed and shows a short local summary of article paragraphs the user scrolled past.“I built a Chrome extension that summarizes the paragraphs you scroll past — Gemma 4 via WebGPU runs locally, nothing is uploaded”
- nano-web-agent — Lets an LLM perceive a web page via its accessibility tree and execute clicks, typing, and scrolling in a perceive-think-act loop.“Built a Chrome extension that lets local/cloud LLMs actually control your browser (click, type, scroll)”
- RouterDash — Compares responses from multiple LLMs side by side on one prompt, showing latency, token counts, and estimated cost.“I made RouterDash — a FREE benchmark playground for OpenRouter LLMs”
- GIDE — Routes coding editor tasks between a local on-device GGUF model for cheap high-frequency work and cloud models like Claude, GPT, or Gemini via the user's own API key for harder reasoning.“A week of real coding with local inference and zero token cost. Our editor routes the cheap stuff to an on-device model”
- MermaidBin — Hosts Mermaid diagrams on shareable pages and stable SVG URLs, and suggests fixes for syntax errors using a local model.“MermaidBin, a pastebin for Mermaid diagrams”
- Glovira — Translates speech in real time on the phone, running speech recognition and a language model locally with cloud fallback.“The Technical Hook”