VSCode + Continue + Gemini

Local config for the Continue VS Code extension, wiring up the full Google Gemini model family through a single config.yaml. Demoed in a YouTube video -- the index.html to-do app in this repo was built live using Continue during that session.

Related YouTube Video

Google AI Studio in VS Code

What's in this repo

File	Purpose
config.yaml	Continue local config with every current Gemini model pre-wired
prompt.txt	The prompt used to generate index.html during the demo
index.html	The resulting to-do app -- zero dependencies, runs in any browser

Setup

1. Get a Gemini API key

Head to Google AI Studio and create a free API key.

2. Install Continue

Search for Continue in the VS Code Extensions marketplace and install it.

3. Drop in the config

Copy config.yaml to your Continue local config directory:

• macOS / Linux: ~/.continue/config.yaml
• Windows: %USERPROFILE%\.continue\config.yaml

Replace every <YOUR-API-KEY> placeholder with your actual key.

4. Reload VS Code

Open the Continue sidebar and the Gemini models will appear in the model picker.

Models included

Model	Notes
Gemini 2.5 Pro	Best reasoning, slower
Gemini 2.5 Flash	Fast, strong default
Gemini 2.5 Flash Lite	Cheapest / fastest
Gemini 2.0 Flash / Flash Lite	Previous generation
Gemini 3.x Flash / Pro Preview	Latest previews
Gemma 4 31B / 26B	Open weights via Gemini API
Deep Research models	Long-context research variants
Antigravity Agent Preview	Experimental agentic model

Demo app

The index.html task manager was generated from prompt.txt in a single Continue chat. Open it directly in a browser -- no build step, no server needed.

Tips

• Use Gemini 2.5 Flash as your everyday model (speed vs. quality sweet spot).
• Switch to Gemini 2.5 Pro for complex refactors or architecture questions.
• The roles field in config.yaml (commented out on Flash Lite) lets you pin a model to autocomplete only -- uncomment it to save quota on heavier models.