Edge Studio
Edge Studio is a local web workbench for model optimization, validation, export, and device coordination. It takes a source model and produces deployment-ready artifacts: Edge Kit bundles, GGUF/CoreML exports, Neural Imprint artifacts, or a complete Edge Scaffold project.
Developer Preview
Validate every exported model on the target device before shipping. Build success alone is not sufficient.
Where it fits
Source model → Edge Studio → Optimized bundle → Edge Kit (inference) → Your agent
Neural Imprint artifact → Edge Halo restore flow
Edge Scaffold project → Xcode → Your agent
Edge Studio is the local workbench. Edge Kit and Edge Halo are the runtime packages. They are separate — your shipping agent does not depend on Edge Studio.
How to launch
python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install edge-studio
edge studio
Open http://127.0.0.1:18842. For frontend development with Vite, see
Install Edge Studio.
Two modes
| Mode | For | Route |
|---|---|---|
| Simple | First-time users. Guided wizard: detect device → pick model → optimize → test → export. | /simple |
| Pro | Full workbench. 20+ tools for analysis, optimization, benchmark, export. | /dashboard |
Capabilities
| Area | Tools |
|---|---|
| Analysis | Architecture browser, weight analysis, activation heatmap, attention patterns, KV cache analysis, MoE analyzer, inference tracer, model comparison |
| Optimization | Advisor, auto optimizer, pipeline, pruning simulator, mixed precision, quality validator, distillation, merge, auto tune |
| Testing | Multi-modal chat (LLM/VLM/STT/TTS), voice duplex |
| Batch | Multi-model benchmark dashboard, batch operations |
| Personalization | Neural Imprint generation, profile artifact inspection, model-matched profile resources, device backflow |
| Devices | EdgeMesh pairing, trusted peer state, capsule push/receive, apply-status receipts |
Developer workflow
- Load or select a source model.
- Inspect model architecture and device fit.
- Optimize and benchmark with the target device in mind.
- Export an Edge Kit bundle or Edge Scaffold project.
- Pair a test device with EdgeMesh when personalization or device receipts are needed.
- Generate or inspect Neural Imprint artifacts locally.
- Validate on a real device before shipping.
Edge Studio prepares and audits artifacts. Your shipping agent consumes those artifacts through Edge Kit and Edge Halo.
Requirements
- Python 3.11+, Node.js 18+
- Recommended: Mac with sufficient memory for the source model