SHARED AI · A FIELD STATION EXPERIMENT DISTRIBUTED INFERENCE

OPEN COMPUTE / PRIVATE DATA: NO Do not enter private or sensitive information. Every participating device helps compute the answer. WebRTC encrypts traffic in transit, but model activations should be treated as visible to room participants.

START · SHARED AI

Create a room and invite other devices.

A protected invitation link will be created automatically. Share that link with the other people or devices you want to include.

GPU CONTRIBUTION GB

BEFORE YOU CONTINUE

This is an experiment, not a private AI service. Prompts are not confidential. Other participating devices help compute the answer and are not independently verified. Joining can use significant GPU, memory, bandwidth and battery.

By creating the room, you acknowledge the notice above.

Room

----

Select the room label to copy the protected invitation link.

Participating devices

Shared model

MODEL / DISTRIBUTED

Measuring room capacity…

Split across the participating devices

Your contribution connected

This device is one part of the shared model.

Shared model

Waiting for model

GPU offered—

Compute completed—

Data exchanged—activation traffic

Tokens processed—

Model data loaded—

Together Waiting for other devices

This device

Measured in this browser. Compute time is model-operation time; data is activation traffic over the FIELD STATION wire. Battery appears only when the browser reports it. Hardware temperature and true GPU utilisation are not available to the page.

CONTEXT / new room conversation

Clear cached model weights

ROOM / —

Connect enough devices, choose a model, then start the experiment.

MODEL DISTRIBUTION /

Each device fetches only the model layers assigned to it. Download rate and request count are shown while loading.