Studio UI reference
A quick reference for every route in the Edge Studio web interface. For the main workflow, see Studio overview and Optimize and benchmark.
Simple mode
| Route | Page | What it does |
|---|---|---|
/simple | Device profile | Auto-detect hardware, show AI capability rating |
/simple/focus | Focus selection | Choose: Chat, Coding, Vision, ASR, TTS, or Voice Duplex |
/simple/tier | Tier selection | Pick model size: Standard / Pro / Max / Ultra |
/simple/setup | Setup | Download model, load, test with embedded panel |
/simple/done | Complete | Success state, path to export |
/simple/export/device | Export device | Choose iPhone or iPad target |
/simple/export/generate | Export generate | Device-fit check, app generation |
Pro mode — Dashboard
| Route | Page | What it does |
|---|---|---|
/dashboard | Pro dashboard | Stage-aware cockpit: loaded → profiled → optimized → exported |
Pro mode — Analysis
| Route | Page | What it does |
|---|---|---|
/architecture | Architecture browser | Layer tree, parameter counts, 3D view |
/weights | Weight analysis | Tensor sizes, distributions, outliers |
/activation | Activation heatmap | Per-layer activation magnitudes |
/attention | Attention patterns | Head importance, matrix views |
/kv-cache | KV cache analysis | Memory projections, device-fit table |
/moe | MoE analyzer | Expert routing and load balance |
/inference | Inference tracer | Token probabilities, step-by-step |
/comparison | Model comparison | Side-by-side A vs B |
Pro mode — Optimization
| Route | Page | What it does |
|---|---|---|
/optimization | Advisor | Recommended optimization plan |
/auto-optimizer | Auto optimizer | Automated candidate search |
/pipeline | Pipeline | Manual step-by-step optimization |
/pruning | Pruning simulator | Preview size reduction |
/mixed-precision | Mixed precision | Per-layer bit-width control |
/quality | Quality validator | Perplexity, benchmark, custom prompts |
/distill | Distillation | Student model from teacher |
/merge | Merge | Combine compatible model sources |
/auto-tune | Auto tune | Search inference parameters |
Pro mode — Testing
| Route | Page | What it does |
|---|---|---|
/chat | Chat | Multi-modal: LLM / VLM / STT / TTS |
/duplex | Duplex chat | Voice conversation: ASR → LLM → TTS |
Pro mode — Batch
| Route | Page | What it does |
|---|---|---|
/benchmark-dashboard | Benchmark dashboard | Multi-model benchmark, Plotly charts, CSV export |
/batch | Batch operations | Queue multiple models for optimization |
Pro mode — Other
| Route | Page | What it does |
|---|---|---|
/neural-imprint-chat | Neural Imprint Chat | Preview a generated Neural Imprint artifact with a Mac-side base model |
/devices | Devices | EdgeMesh device management and pairing |
/export | Export | Wizard-style export flow |