🚀 LiteLLM Connector for GitHub Copilot Chat

Open VSX Version Open VSX Downloads

Bring any LiteLLM-supported model into the Copilot Chat model picker — OpenAI, Anthropic, Google, Mistral, local Llama, and more. If LiteLLM can talk to it, Copilot can use it.

🆕 What's New in 2.2.2

Version 2.2.2 corrects per-effort reasoning metadata handling so the model picker preserves defaults and applies explicit LiteLLM effort states correctly.

🧠 Per-effort reasoning states — Baseline null or absent fields retain the default effort, false removes only that effort, and true retains it.
🧩 Extended effort support — Explicitly supported minimal, xhigh, and max values are added without replacing the baseline effort list.
🎛️ Field-specific model-card overrides — Overrides continue to change only the exact LiteLLM fields they define.

See CHANGELOG.md for previous release notes.

⭐️ Support the Project

⭐ Star on GitHub: https://github.com/gethnet/litellm-connector-copilot
📝 Leave a review on the VS Code Marketplace
☕ Support development: Ko-fi | Buy Me a Coffee

⚡ Quick Start (60 Seconds)

Install GitHub Copilot Chat (if not already installed)
Install LiteLLM Connector for Copilot
Open Command Palette (Ctrl+Shift+P)
Run LiteLLM: Manage Configuration
Add a provider group:
- Name (e.g., "Cloud", "Local")
- Base URL (e.g., http://localhost:4000)
- API Key (required)
Open Copilot Chat → pick a model → start chatting!

✅ Requirements

🖥️ VS Code 1.120+
🤖 GitHub Copilot (Free Individual plan works)
🌐 A LiteLLM proxy URL and API key

✨ Features & Differentiators

Feature	Why It Matters
🔌 Direct LiteLLM Integration	No third-party wrappers — talks to your proxy directly with native message formatting, streaming, and tool handling
🧩 Native VS Code Integration	Model picker groupings, category tags, reasoning effort selectors, token indicators — all first-class in VS Code's Language Model API
👤 Single-Maintainer Project	Direct access to the person who builds it. Fast decisions, straightforward communication. We test thoroughly but things slip through — report issues, we respond.
🌍 Any Model	Access GPT-4, Claude, Gemini, Llama, DeepSeek, and more
⛓️ Multi-Backend	Aggregate from multiple proxies with proper isolation — each backend stays grouped in the picker
💭 Thinking Support	Full Anthropic thinking content (signatures, redacted, display metadata)
🌊 Real-Time Streaming	Watch responses as they're generated
🛠️ Tool Calling	Models can use tools to interact with your workspace
👁️ Vision	Image analysis support
📊 Token Tracking	Real-time input/output token usage
✍️ Commit Generation	Generate conventional commit messages from staged changes
🔐 Secure	API keys stored in VS Code's encrypted storage

🐛 Troubleshooting

Models not showing up?

Run LiteLLM: Manage Configuration and verify Base URL + API key
Run LiteLLM: Reload Models to force refresh
If stuck: Remove LiteLLM provider groups via LiteLLM: Manage Configuration → VS Code's Language Models UI, then re-add

⚙️ Configuration

Base URL + API key are configured through VS Code's Language Models UI (run LiteLLM: Manage Configuration).

Standard Settings

Setting	Default	Description
`commitModelIdOverride`	`""`	Model ID for commit message generation
`inactivityTimeout`	`60`	Seconds before stream is considered idle
`disableCaching`	`false`	When enabled, bypass LiteLLM caching for models that advertise support for the `cache` parameter
`enableModelOverrides`	`false`	Enable model-card override rules
`displayPricingInPicker`	`true`	Show model pricing in picker
`discoveryTimeoutMs`	`5000`	Timeout (ms) for model discovery
`discoveryCacheTtlMs`	`60000`	Cache TTL (ms), 0 to disable
`discoveryFireDebounceMs`	`250`	Debounce (ms) for change notifications
`discoveryFireMinIntervalMs`	`2000`	Min interval (ms) between notifications

Reasoning model-card overrides are disabled by default. Enable enableModelOverrides when LiteLLM reports incorrect or incomplete reasoning metadata. Overrides replace or add only the explicitly named LiteLLM fields; related fields are not inferred.

🛠️ Help: Applying a Model Override

Model overrides are disabled by default. To correct incomplete LiteLLM /model/info metadata:

Open Preferences: Open User Settings (JSON) or Preferences: Open Workspace Settings (JSON).
Set litellm-connector.enableModelOverrides to true.
Add a matching rule to litellm-connector.modelOverrides.
Run LiteLLM: Reload Models.

Use the raw LiteLLM model_name and exact snake_case model-card fields. Only explicitly defined fields are changed; related fields are not inferred.

{
   "litellm-connector.enableModelOverrides": true,
   "litellm-connector.modelOverrides": [
      {
         "match": "^gpt-4\\.8$",
         "supports_reasoning": true,
         "supports_max_reasoning_effort": true
      }
   ]
}

Define each desired effort explicitly, such as supports_xhigh_reasoning_effort: true. Setting one effort field does not enable supports_reasoning or any other effort field automatically.

Advanced (JSON-Only)

These aren't in Settings UI — add to settings.json if needed:

Setting	Default	Why Use It
`forceResponsesEndpoint`	`false`	Force all models to use `/responses` endpoint for consistent reasoning/thinking support
`allowChatCompletionsFallback`	`false`	Fall back to `/chat/completions` if `/responses` fails (needs `forceResponsesEndpoint: true`)

⌨️ Commands

LiteLLM: Manage Configuration — Add/edit provider groups
LiteLLM: Reload Models — Refresh model list
LiteLLM: Show Available Models — View discovered models
LiteLLM: Generate Commit Message — Generate commit from staged changes
LiteLLM: Set Log Level — Change logging verbosity

LiteLLM Connector for Copilot

Gethnet