Get startedLeaderboardDecode calculatorModelsReportsHardwareBenchmarksMarketplaceRentalsProAPI Docs
Language
Total runs
Highest
Best memory ceiling
Median
Lowest
NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 — 285.6 tok/s on NVIDIA GeForce RTX 5090 · 32 GB | Localmaxxing