View source for Code with local AI, linux & VSCode

<ul>
<li><p>install ollama: https://www.ollama.com/download/linux</p></li>
<li><p>how to download a model from https://www.ollama.com/search</p>
<ul>
<li><p>choose gpt-oss, deepseek-r1, gemma or qwen3-coder</p></li>
<li><p>type in the terminal “ollama pull gpt-oss:20b”</p></li></ul>
</li>
<li><p>install “continue” extension in VSCode https://docs.continue.dev/getting-started/install</p></li>
<li><p>drag the “continue” extenstion icon from the left to the right side</p></li>
<li><p>configure the extension:</p>
<ul>
<li><p>there is no need to login</p></li>
<li><p>in the “select model” under the chat, select “Ollama” and the installed model</p></li>
<li><p>if you ask anything, it will throw an error that the model can’t be found</p></li>
<li><p>In the config.yaml that should have been opened (/home/$user/.continue/config.yaml) modify “models” section for this:</p>
<syntaxhighlight lang="yaml">models:
  - name: gpt-oss:20b
    provider: ollama
    model: gpt-oss:20b
    roles:
      - chat
      - edit
      - apply
      - autocomplete</syntaxhighlight>
<p>This is because there is some kind of bug (right now) and it appends “openai” in the model name, it doesn’t use the proper name we pulled before from ollama repository.</p></li></ul>
</li>
<li><p>add another models depending on your hardware capabilities:</p>
<ul>
<li><p>RTX 3090</p>
<ul>
<li><p>gpt-oss:20b</p></li>
<li><p>deepseek-r1:14b</p></li>
<li><p>qwen3-coder:30b</p></li>
<li><p>llama3.1:8b</p></li></ul>
</li></ul>
</li></ul>

[[Category:AI Resources]]