NewChunking Qwen3.5's gated DeltaNet for 1.5x faster prefill on Apple Silicon
Monitor

See every inference your fleet runs.

Trace every inference in the field. Catch drift before it becomes a problem.

See it live →

Models fail quietly in the field.

Trace every inference

Inputs, outputs, timings, and device state, tied to the exact build that produced them.

Metrics per build

Track throughput, latency, power, and accuracy per device and per version.

Debug without the SSH session

Pull any run off any device and replay it at your desk, exactly as it happened.

From one frame to the whole fleet.

End-to-end traces

Inputs, outputs, and timings for every inference, tied to the build that produced them.

Latency breakdowns

See percentiles per device and per version, split between the accelerator and postprocessing.

Drift alerts

A drop in accuracy flags the affected devices automatically.

Search and filter

Filter runs by device, model, version, or fault to move from symptom to cause.

Replay

Replay any inference exactly as the device ran it.

Inspect from the terminal

Pull up any run from the command line: inputs, timings, and everything the vendor runtime printed.

See everything your fleet does.

Open the fleet view and watch the first runs come in.