The leading open source AI code assistant. You can connect any model and any context to build custom autocompletion and chat experiences within the IDE!
The following open source models are recommended:
Chat: llama3.1-8B
Reasoning code: deepseek-coder-v2:16b
embedding model
nomic-embed-text
Default storage path for models: C:\Users\your username\.ollama\models\blobs
Model offline download: /s/dbc3fdeffc68
Command Line Run:
ollama run deepseek-coder-v2:16b
ollama run llama3.1:8b
ollama pull nomic-embed-text:latest
VScode Search: Continue
Example of a dialog model:
"models": [
{
"title": "Ollama",
"provider": "ollama",
"model": "llama3.1:8b"
}
],
"models": [
{
"model": "gpt-4-turbo-preview",
"contextLength": 8192,
"title": "gpt-4-turbo-preview",
"systemMessage": "You are an expert software developer. You give helpful and concise responses.",
"apiKey": "you-key",
"apiBase": "https://api./v1",
"completionOptions": {
"temperature": 0.7,
"topP": 0.9,
"topK": 10,
"presencePenalty": 0.01,
"frequencyPenalty": 0.02
},
"provider": "openai"
}
],
Set up code reasoning:
"tabAutocompleteModel": {
"title": "deepseek-coder-v2:16b",
"provider": "ollama",
"model": "deepseek-coder-v2:16b"
},
Setting Embedding:
"embeddingsProvider": {
"provider": "ollama",
"model": "nomic-embed-text"
},
Use the tutorial:
Easy to understand code
VS code:
cmd+L
(MacOS) / ctrl+L
(Windows)Tab autocomplete code
Refactor the function you are coding
VS code:
cmd+I
(MacOS) / ctrl+I
(Windows)Ask a question about your code base
VS Code:
@codebase
(MacOS) / @codebase
(Windows)Quickly use documents as context
VS code:
@docs
(MacOS) / @docs
(Windows)Use the slash command to start the operation
hereLearn more about the slash command
VS Code:
/edit
(MacOS) / /edit
(Windows)Adding classes, files, etc. to a context
VS code:
@files
(MacOS) / @files
(Windows)
hereLearn more about the Context Provider program.
Understanding Terminal Errors
VS code:
cmd+shift+R
(MacOS) / ctrl+shift+R
(Windows)Official Documentation:
/continuedev/continue
/setup/overview
/blog/continue-code-assistant
/setup/select-model