Settings

Configure once — defaults apply across all tools automatically.

Default model

Pre-filled in Predict performance, KV cache calculator, and Recommend sizing. Use any Hugging Face model ID.

Model — Hugging Face ID

All tools use this model by default. Type to autocomplete from the catalog.

Tested models▼

Models tested for use with the AISimulators sizing engine.

0 models

Hugging Face token

Required for gated or private models. Stored in this browser only — never sent to our servers.

API token

Simulator

Simulator used by Predict performance and Recommend sizing.

Simulator

Predict and Recommend use this simulator. BLIS is available with vLLM; KV cache uses AISimulate.

Inference backend

Serving framework targeted when generating configuration recommendations.

Backend

Version