Install Edge Studio
Create and activate a Python 3.11 environment, then install Edge Studio from the Python package for normal local development:
python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install --upgrade --pre edge-studio
edge doctor
If you use uv:
uv venv --python 3.11 .venv
source .venv/bin/activate
uv pip install --upgrade --pre edge-studio
edge doctor
--pre installs the current Developer Preview release candidate. Keep it until
the first stable package is published. edge doctor checks the Python
environment, model paths, and system compatibility; fix any failed checks before
continuing.
Requirements
| Requirement | Version |
|---|---|
| macOS | 14 or later |
| Hardware | Apple Silicon |
| Python | 3.11 recommended |
| Node.js | Not required for the packaged Studio UI |
Use edge for model readiness checks, model fetch receipts, and local learning demos:
edge models where qwen3.5-9b-4bit --json
edge models fetch qwen3.5-9b-4bit --source auto
edge demo chat --model qwen3.5-9b-4bit --interactive
edge models fetch --source auto can select the best available preview download path from ModelScope, Hugging Face, or an HF mirror. The download is explicit and writes a receipt.
The qwen3.5-9b-4bit download is approximately 5 GB, and the time depends on
your network.
After [chat:ready], ask a few normal questions and exit with /exit. The first 9B model load can take tens of seconds.
After the base chat works, continue to Build your first device Agent to inspect the synthetic finance signal and see user-specific learning restored without creating a new model release. Then export the same lifecycle into an Agent carrier.
Launch the Web UI
The installed package exposes a single edge command. Start the local Studio UI
and API server with:
edge studio
Open:
http://127.0.0.1:18842
The server runs on localhost by default. Stop it with Ctrl+C.