Local models
26 entries.
Runtimes and code models you can run on your own machine, without sending the repo out.
- All
- Editors
- CLI agents
- Cloud agents
- Code review
- App builders
- Local models
- Picks first
- A to Z
- By category
26
-
NVIDIA's desktop demo app for running local models on GeForce RTX GPUs, including some coding use.
Windows
-
Meta's open code-model family, still a common local baseline even as newer models have passed it on many benches.
Local
-
Google's open Gemma variant trained for code completion and generation, meant to be run locally or through a host.
Local
-
Mistral's family of models trained for code, available through Mistral's API and as weights you can run locally.
API, Local
-
DeepSeek's code-model family, including open weights that people run locally and the hosted models used by many coding agents.
Local, API
-
Mistral's agentic coding model family, aimed at repository tasks and available to run or call through Mistral.
API, Local
-
Docker's local model runner, so a coding tool can call a model packaged and served through Docker Desktop.
macOS, Windows, Linux
-
Microsoft's local runtime for running foundry-hosted models on a Windows or Mac machine, with an API that tools can call.
Windows, macOS
-
Nomic's desktop app and runtime for running local models with a simple installer, including a local API.
macOS, Windows, Linux
-
IBM's open Granite models for code, released for local and enterprise use with a relatively permissive licence.
Local, API
-
An open-source desktop app for running local models offline, with a ChatGPT-like interface and an OpenAI-compatible local server.
macOS, Windows, Linux
-
The widely used C/C++ runtime for running Llama-family and other models locally, including on CPUs and Apple silicon.
macOS, Linux, Windows
-
Mozilla's single-file format for distributing and running a local model: one executable that starts a server on your machine.
macOS, Linux, Windows
-
A desktop app for downloading and running local models, with a chat UI and a local server that coding tools can point at.
macOS, Windows, Linux
Editor's pick
-
A self-hosted, OpenAI-compatible API that can serve local models, including for coding tools that expect an API endpoint.
Linux, Docker
-
Apple's array framework for running models efficiently on Apple silicon, with mlx-lm used to serve local code models on a Mac.
macOS
-
The most common way to run open models on a laptop, including several code models, from a single command-line tool and a local API.
macOS, Linux, Windows
Editor's pick
-
A self-hosted chat UI that talks to Ollama and other local runtimes, often used as the front end for a private code model.
Self-hosted
-
Alibaba's open code-model family (Qwen2.5-Coder and later), widely run locally through Ollama, LM Studio and llama.cpp.
Local, API
-
A fast open-source runtime for serving large models, used as a local or private backend for agents.
Linux
-
An open code model from the BigCode project (ServiceNow and Hugging Face), trained on The Stack and released for local use.
Local
-
Hugging Face's server for deploying models, often used to host a private code model for an editor or agent.
Linux
-
A widely used local web UI for running and serving models, including code models, on your own hardware.
Windows, Linux, macOS
-
A toolkit for fine-tuning local models quickly, including code models, on a single GPU.
Linux, Windows
-
A high-throughput runtime for serving large models on your own GPUs, often used as the local or private backend for coding agents.
Linux
-
01.AI's open code-model family, available as weights for local inference.
Local
Nothing matches that. Try another word, or suggest a tool.
Suggest a tool
If an AI coding tool is missing, send the name and a link. We only add it after checking the official page.
A note when the list grows
Occasional email when we add a batch of tools worth opening. No ads, and we never sell the address.