🏆
Rankings
LLM Leaderboard
Real measurements of every tested language model on real, named hardware – rated by raw speed (performance in tokens per second, prefill and time to first token) and by practical task quality in complete agent and chat runs (harness). Pick a benchmark type below or filter by model, maker and hardware to see exactly what is tested and how the results are produced.
⚡ Performance (tok/s)🤖 Harness quality👥 Concurrency🖥️ real hardware
| # | Model / Maker | Metrics | Parallel | GPU / CPU / RAM | Runtime | ||
|---|---|---|---|---|---|---|---|
| 1 | Nemotron-3-Nano-4B4BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 367,75 tok/s TG Prefill 3.397 · TTFT 29.121 ms | 10× | NVIDIA GB10 (DGX Spark)NVIDIA Grace · 120 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 2 | Nemotron-3-Nano-4B4BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 354,29 tok/s TG Prefill 3.544 · TTFT 30.045 ms | 10× | NVIDIA GB10 (DGX Spark)NVIDIA Grace · 120 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 3 | Nemotron-3-Nano-4B4BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 210,25 tok/s TG Prefill 3.483 · TTFT 8.798 ms | 5× | NVIDIA GB10 (DGX Spark)NVIDIA Grace · 120 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 4 | Nemotron-3-Nano-4B4BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 202,30 tok/s TG Prefill 3.149 · TTFT 9.380 ms | 5× | NVIDIA GB10 (DGX Spark)NVIDIA Grace · 120 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 5 | Nemotron-3-Nano-4B4BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 76,87 tok/s TG Prefill 2.223 · TTFT 1.006 ms | 1× | NVIDIA GB10 (DGX Spark)NVIDIA Grace · 120 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 6 | Nemotron-3-Nano-4B4BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 72,70 tok/s TG Prefill 1.798 · TTFT 1.470 ms | 1× | NVIDIA GB10 (DGX Spark)NVIDIA Grace · 120 GB RAM | llama.cppgodclawQ4_K_M | Details → |