№ 0934GitHub
InferHub models in GitHub Copilot Chat
A BYOK VS Code extension that registers the InferHub catalog, including Muse Spark with its 1,048,576-token context window, as a model provider in the Copilot Chat picker without Copilot Pro.
InferHub Copilot Chat
BYOK InferHub models in GitHub Copilot Chat. No Copilot Pro needed.
One sk-airo- key gets you the full InferHub marketplace: Muse Spark with a 1,048,576 token context window, Claude, GPT, GLM, Kimi, DeepSeek and friends, routed cheapest-first. This extension registers the catalog as a language model provider so it shows up in the Copilot Chat model picker next to GPT and Claude.
Quick start
- Install GitHub Copilot Chat (required) and this extension.
- Open Copilot Chat.
- Click the model picker, then Manage Models.
- Select InferHub, accept the group name, and paste your
sk-airo-API key. - Pick a model and chat.
Get an API key from InferHub.
Models
The extension registers every model the InferHub catalog returns: concrete rails (ag/, cc/, cmc/, cx/, ocg/, zai/, cb/, cbcn/, cp/, mimo/, ali/) and auto-route aliases. If the catalog is unreachable it falls back to the bundled Muse Spark entries.
Picker names are tagged with the serving provider, e.g.:
| Catalog id | Picker name |
|---|---|
ag/claude-opus-4-6-thinking |
Claude Opus 4.6 (Antigravity) |
cmc/meta/muse-spark-1.2 |
Muse Spark 1.2 (Command Code) |
cx/gpt-5.6-luna |
GPT 5.6 Luna (OpenAI Codex) |
muse-spark-1.2 |
Muse Spark 1.2 (alias) |
Aliases (bare names, no prefix) auto-route to the cheapest provider behind the name — (alias) in the picker, (auto-routed) in the detail line.
Muse Spark keeps its 1,048,576 token context window. The contributor tier is cheaper because the upstream trains on your prompts and completions; the standard tier does not.
Endpoint
OpenAI-compatible responses and chat completions at https://api.inferhub.dev/v1. Tool calling works, so agent mode functions normally.
Usage
Run InferHub: Show InferHub Usage for a breakdown of the last 24 hours straight from the InferHub management API: per-provider token/cost/latency/speed tables plus per-request rows (status, tokens, cost, TTFT, duration, tok/s).
The status bar also shows a per-response prompt/output token summary with cost when the API reports it.
Settings
| Setting | Default | Description |
|---|---|---|
inferhub.baseUrl |
(empty) | Override the API base URL |
inferhub.temperature |
0.2 |
Sampli |
