ModelBench
Which model wins on your workload?
ModelBench answers which model to use by measuring, not guessing. Build prompt suites (or use the seeded reasoning / coding / summarization sets), pick the models to compare, and each prompt runs against each model on the live Nyquest relay.
Latency is wall-clock measured, tokens come from the relay usage, cost is computed from the price table, and an optional LLM judge scores quality — clearly labelled as a judge model's opinion. The leaderboard ranks fastest, cheapest, and highest-quality; a failed call is a failure, never hidden.
A Pro workspace gated behind Nyquest Pro, enforced frontend and backend. Export a portable Nyquest Benchmark Report.
Live now. Requires a Nyquest Pro subscription. Every number is measured from a real relay call.