🏆
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 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 62,08 tok/s TG Prefill 1.013 · TTFT 161.953 ms | 10× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 2 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 61,28 tok/s TG Prefill 761 · TTFT 168.314 ms | 10× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 3 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 32,07 tok/s TG Prefill 958 · TTFT 47.870 ms | 5× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 4 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 31,63 tok/s TG Prefill 812 · TTFT 50.633 ms | 5× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 5 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 9,40 tok/s TG Prefill 644 · TTFT 3.793 ms | 1× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 6 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 9,40 tok/s TG Prefill 547 · TTFT 4.475 ms | 1× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 7 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 2,67 tok/s TG Prefill 78 · TTFT 65.710 ms | 10× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppopenclaw_cli batched-bench ROCmQ4_K_M | Details → | |
| 8 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 2,16 tok/s TG Prefill 78 · TTFT 32.868 ms | 5× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppopenclaw_cli batched-bench ROCmQ4_K_M | Details → | |
| 9 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,68 tok/s TG Prefill 48 · TTFT 183.808 ms | 10× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 10 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,63 tok/s TG Prefill 51 · TTFT 181.415 ms | 5× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 11 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,96 tok/s TG Prefill 48 · TTFT 187.060 ms | 10× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 12 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,94 tok/s TG Prefill 51 · TTFT 180.358 ms | 5× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 13 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,93 tok/s TG Prefill 51 · TTFT 179.975 ms | 5× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 14 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,81 tok/s TG Prefill 4 · TTFT 535.235 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 15 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,81 tok/s TG Prefill 4 · TTFT 534.264 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 16 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,70 tok/s TG Prefill 75 · TTFT 6.811 ms | 1× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppopenclaw_cli batched-bench ROCmQ4_K_M | Details → | |
| 17 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,67 tok/s TG Prefill 35 · TTFT 60.617 ms | 1× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 18 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,65 tok/s TG Prefill 4 · TTFT 485.276 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 19 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,65 tok/s TG Prefill 4 · TTFT 482.771 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 20 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,65 tok/s TG Prefill 4 · TTFT 488.087 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 21 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,63 tok/s TG Prefill 4 · TTFT 479.913 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 22 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,63 tok/s TG Prefill 4 · TTFT 477.862 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 23 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,63 tok/s TG Prefill 4 · TTFT 521.790 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 24 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,62 tok/s TG Prefill 4 · TTFT 481.185 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 25 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,61 tok/s TG Prefill 4 · TTFT 483.799 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 26 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,57 tok/s TG Prefill 4 · TTFT 531.701 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 27 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,55 tok/s TG Prefill 4 · TTFT 527.248 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 28 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,54 tok/s TG Prefill 4 · TTFT 531.922 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 29 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,51 tok/s TG Prefill 4 · TTFT 525.090 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 30 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,29 tok/s TG Prefill 36 · TTFT 58.861 ms | 1× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 31 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,28 tok/s TG Prefill 30 · TTFT 74.267 ms | 1× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 32 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,11 tok/s TG Prefill 8 · TTFT 281.154 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 33 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,10 tok/s TG Prefill 7 · TTFT 320.464 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 34 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,10 tok/s TG Prefill 9 · TTFT 247.886 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 35 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,10 tok/s TG Prefill 8 · TTFT 272.289 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 36 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,10 tok/s TG Prefill 7 · TTFT 257.748 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 37 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,10 tok/s TG Prefill 8 · TTFT 263.746 ms | 1× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 38 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 7 · TTFT 304.304 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 39 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 8 · TTFT 275.212 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 40 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 8 · TTFT 265.674 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 41 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 7 · TTFT 328.995 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 42 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 8 · TTFT 273.694 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 43 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 7 · TTFT 319.557 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 44 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 8 · TTFT 280.941 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 45 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 7 · TTFT 296.158 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 46 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 7 · TTFT 289.112 ms | 5× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 47 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 8 · TTFT 280.190 ms | 1× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 48 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,09 tok/s TG Prefill 8 · TTFT 260.023 ms | 1× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 49 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,08 tok/s TG Prefill 8 · TTFT 280.103 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 50 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,08 tok/s TG Prefill 7 · TTFT 303.658 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 51 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,08 tok/s TG Prefill 7 · TTFT 309.703 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 52 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,08 tok/s TG Prefill 7 · TTFT 327.762 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 53 | Llama-3.1-Nemotron-Ultra-253B-v1253BNVIDIA Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,05 tok/s TG Prefill 8 · TTFT 290.462 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → |