Slice your prompt into labeled layers and watch each one eat into the model's context window — built on the Reactor developer platform.

Idle
Tokens produced
0 / 500 tokens assigned
Layers

No layers yet. Click + Add layer, then highlight prompt regions while the layer is active.