llama.cpp (chat or unspecified)
How vibe coders and indie builders use llama.cpp (chat or unspecified): 22 projects built with it since May 2026.
Projects built with it
22
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 6
- Qwen 6
- Claude API 4
- Ollama 4
- Gemma 4
- Rust 3
- Gemini API 3
- Metal 3
What's built with it
Latest projects built with llama.cpp (chat or unspecified)
- Hefesto — Turns an Android phone into a development workspace that compiles APKs on-device and runs coding agents inside a Linux environment.“Have you ever thought about building apps and websites right from your phone?”
- Pricacy-first local AI… — Runs local LLMs as a chat and document review assistant that analyzes PDFs, Office files, and images without sending data to the cloud.“Pricacy-first local AI interface for document reviews and chat”
- ramdeck-core — Pools RAM and VRAM across networked machines to run a large language model behind an OpenAI-compatible API.“Pooled RAM across an old laptop, a Windows PC, and a Mac to run a 13B model - source-available, would love eyes on it”
- Avatar.gift — Runs an on-device AI assistant on an iPhone that handles reminders, notes, messages, trackers and memory without a server.“Spent the last 6 months building an AI that lives on your iPhone and slowly becomes a copy of you. No server, no account. It's out now.”
- amble — Turns a sentence describing the user's day into a structured schedule using a small on-device language model.“The scariest line in my SaaS spreadsheet was the OpenAI bill, so I moved the model on-device.”
- Lexo — Provides an Android keyboard with on-device LLM-blended word suggestions, autocorrect, glide typing, and offline speech dictation.“I got tired of Gboard's mediocre suggestions and dictation so I built an Android keyboard that runs speech and language models entirely on your phone”
- Glovira — Translates speech in real time on the phone, running speech recognition and a language model locally with cloud fallback.“The Technical Hook”
- Minnarone — Runs multimodal agents that take live chat, audio, and video as input and generate contextual reactions from them.“Minnarone - Multimodal agents that watch, listen, and react to live context”
- Tellama — Runs local GGUF language models fully on an Android phone and exposes an Ollama/OpenAI-compatible HTTP API.“I built Tellama, an Android app that runs local LLMs entirely on-device (with an Ollama/OpenAI-compatible API server)”
- EdgeLM — Runs a shared on-device language model on Android that other apps can call over Binder to stream chat tokens.“EdgeLM — a shared on-device AI runtime any app can call via EdgeLM.chat() (Binder service + llama.cpp)”
- Fluent AI — Runs large language models on an Android phone for offline chat and PDF question answering, with optional cloud model access.“Fluent AI — offline AI chat on Android”
- Feral — Runs local GGUF models and cloud AI models in a desktop workspace with sandboxed agent tools, persistent memory graph, and MCP server installation.“Feral v0.2.0 - open-source local AI workspace (llama.cpp + BYOK + agent runtime), now on Windows, macOS and Linux. No telemetry, no subscription, MIT/Apache-2.0”
- Axiom — Runs a local multi-model AI council that plans, executes, and reviews tasks, alongside a chat app that executes code inline.“I spent months building a free Windows AI app, with an AI council mode. no subscription, no account, no data leaving your machine”
- Cipher AI — Runs LLMs on an iPhone offline and connects the phone to models served from a user's own Mac or PC.“Built an iPhone app that runs LLMs offline on the device and can also connect to Ollama/LM Studio on your own computer”
- Sombra — Predicts text continuations and spelling corrections in any macOS text field using a locally run language model, accepted with the Tab key.“I built a small local-AI autocomplete for macOS (personal project)”