Local and open models

Running models on your own hardware keeps prompts and code inside your network. That matters for teams with strict rules about source code, for developers who work offline, and for anyone who wants to experiment with open code models without paying per request. The tools here make it practical: Ollama and LM Studio download and serve models on a laptop or workstation, llama.cpp is the engine many local tools are built on, and Hugging Face is where most open models are published.

For a team, Tabby goes a step further: it is a self-hosted coding assistant that gives everyone code completion and chat from models running on the company's own servers. Editor extensions such as Continue and terminal tools such as Aider can also point at a local model served by Ollama or LM Studio, so you keep the workflow and change only where the model runs.

The trade-off is capability and hardware. Models small enough to run on a laptop are useful for completion, explanations and small edits, but usually trail the largest hosted models on hard reasoning tasks. Larger models need serious GPU memory. Test the models you are considering on your own code before deciding.

Check licenses as well. Open model licenses differ, and some limit commercial use or require attribution. Read the license on the model page before building it into a product or a company workflow.

Featured AI coding tools: local and open models

SponsoredLocal and open models This position is open Put your tool at the top of the local and open models page. Paid spots are always labeled Sponsored. From $99 a month. Get featured

What a featured listing costs ยท from $99 a month

5 AI coding tools in local and open models

Ollama

Runs open models on your own computer with one command and serves them through a local API that editors can use.

  • Free plan or free version
  • Open source
  • Can run on your own servers or machine

LM Studio

Desktop app for finding, downloading and chatting with open models locally, with a local server other tools can call.

  • Free plan or free version
  • Can run on your own servers or machine

Tabby

by TabbyML

Open-source, self-hosted coding assistant that gives a team code completion and chat on its own servers.

  • Open source
  • Can run on your own servers or machine
  • Works in VS Code

Hugging Face

Hub for open models and datasets, where most open code models are published, with tools to run and deploy them.

  • Free plan or free version

llama.cpp

by ggml

Open-source C/C++ engine for running open models efficiently on ordinary hardware, used inside many local AI tools.

  • Free plan or free version
  • Open source
  • Can run on your own servers or machine

A sensible setup for many teams is mixed: local or self-hosted models for completion and anything involving sensitive code, and an approved hosted model for the occasional hard problem. Tools such as Continue and Aider make that easy by letting you switch models per task.

Plan for upkeep. New open models appear often, and swapping one in is easy, but each change can shift quality in ways that are hard to see. Keep a small set of real tasks to test any new model against before you change the team default.

Security still applies on your own hardware. A local model server listening on the network can be reached by anything else on that network unless you restrict it, and downloaded model files should come from sources you trust. Keep the serving tools updated, bind them to localhost on laptops, and put team servers behind the same access controls as other internal services.

Questions about local and open models

What hardware do I need to run a coding model locally?

It depends on the model size. Small models run on a recent laptop, especially with a GPU or Apple silicon. Larger models need much more memory. Model pages usually list the requirements.

Are local models as good as hosted ones for coding?

Usually not for the hardest tasks, but they are good enough for completion, explanations and routine edits. Try them on your own work and keep a hosted model for the difficult jobs if your rules allow it.

Can I use a local model inside my editor?

Yes. Extensions such as Continue, and self-hosted assistants such as Tabby, connect to local or self-hosted models. Ollama and LM Studio expose a local API these tools can call.

Are open models free to use at work?

Not always. Each model has its own license, and some restrict commercial use or require attribution. Read the license on the model page before using it for company work.

Related categories