FlywheelApp · full
Cinema + labs + WebGL. The labeled map, Hub canvas, field sketch, and every live engine. Use when this is the app that owns the GPU on the machine.Every feature, no graphics. Same engines, same API, same honesty grammar. No WebGL, no custom cursor, no Three.js. For when other apps already hold the GPU, or after learning loops when you only need the work.Terminal + one API out. Every live verb as a command and as HTTP. TUCE agents load skill.md and call /api/v1. No cinema overhead. Occupancy stays nominated | live-bound | unavailable.

Lab · 01 · RAM · weights · KV

Fit planner

Illustrative fit from this-tab memory, weight class, context, and LoRA rank. Accelerators stay unavailable. Not POST /api/system/scan.

NCM 1.19.0 · Hub 2.2.0 · invented_metrics = false

illustrative

This-tab hint

Accelerators stay unavailable.

deviceMemory

unavailable

Browser hint, not Hub RAM, not VRAM.

hardwareConcurrency

1

Logical threads reported by this tab.

accelerators

unavailable

No invented VRAM. No vendor GPU shop. No silhouettes.

RAM-class picker

Not a shop. Pick a class, watch the matrix.

Classes are illustrative GiB buckets. Fleet stays unavailable as a fit hint rather than inventing a cluster. Tokens/sec are not published.

Planner

Weights + KV + one LoRA. Architecture knobs are live.

weights

3.73 GiB

KV

1.00 GiB

LoRA

0.0313 GiB

unavailable

4.76 GiB hot

Accelerators stay unavailable. This-tab deviceMemory is a browser hint, not Hub RAM, not VRAM. No RAM-class hint — fit stays unavailable rather than invented. LoRA uses the shared formula 2·r·h·L·M FP16 bytes (4 modules). compose=single. Hint used: unavailable.

classhot GiBverdict on this band
1B-class Q40.65unavailable
3B-class Q42.23unavailable
7B-class Q44.38unavailable
8B-class Q44.76unavailable
8B-class Q88.48unavailable
14B-class Q47.82unavailable
32B-class Q416.98unavailable
70B-class Q435.25unavailable