🏆
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 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 143,05 tok/s TG Prefill 1.609 · TTFT 71.138 ms | 10× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 2 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 123,54 tok/s TG Prefill 2.199 · TTFT 79.387 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 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 122,08 tok/s TG Prefill 1.589 · TTFT 83.232 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 → | |
| 4 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 95,87 tok/s TG Prefill 1.165 · TTFT 108.105 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 → | |
| 5 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 74,09 tok/s TG Prefill 1.590 · TTFT 21.529 ms | 5× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 6 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 64,40 tok/s TG Prefill 1.968 · TTFT 23.096 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 → | |
| 7 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 63,74 tok/s TG Prefill 1.649 · TTFT 24.427 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 → | |
| 8 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 50,80 tok/s TG Prefill 1.092 · TTFT 32.891 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 → | |
| 9 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 50,03 tok/s TG Prefill 564 · TTFT 9.076 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 → | |
| 10 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 36,93 tok/s TG Prefill 619 · TTFT 39.268 ms | 10× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 11 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 30,49 tok/s TG Prefill 1.290 · TTFT 23.886 ms | 10× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 12 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 30,41 tok/s TG Prefill 1.310 · TTFT 23.793 ms | 10× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 13 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 19,71 tok/s TG Prefill 1.011 · TTFT 2.126 ms | 1× | NVIDIA RTX PRO 6000 Blackwell Workstation EditionAMD Ryzen 9 9950X 16-Core Processor · 92 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 14 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 18,76 tok/s TG Prefill 494 · TTFT 5.185 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 → | |
| 15 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 18,55 tok/s TG Prefill 1.165 · TTFT 1.835 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 → | |
| 16 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 18,50 tok/s TG Prefill 980 · TTFT 2.182 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 → | |
| 17 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 17,44 tok/s TG Prefill 894 · TTFT 14.722 ms | 5× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 18 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 15,83 tok/s TG Prefill 699 · TTFT 3.073 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 → | |
| 19 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 15,68 tok/s TG Prefill 1.025 · TTFT 12.644 ms | 5× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 20 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 15,21 tok/s TG Prefill 1.051 · TTFT 12.339 ms | 5× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 21 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 7,19 tok/s TG Prefill 276 · TTFT 1.853 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 → | |
| 22 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 7,03 tok/s TG Prefill 480 · TTFT 3.968 ms | 1× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 23 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 6,94 tok/s TG Prefill 438 · TTFT 4.915 ms | 1× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 24 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 6,83 tok/s TG Prefill 476 · TTFT 4.001 ms | 1× | 4x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 25 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 6,50 tok/s TG Prefill 323 · TTFT 6.557 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 26 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmark | 6,50 tok/s TG Prefill 323 · TTFT 6.557 ms | 1× | 3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | godclawQ4_K_M | Details → | |
| 27 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 4,54 tok/s TG Prefill 143 · TTFT 70.040 ms | 10× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 28 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 4,47 tok/s TG Prefill 143 · TTFT 69.846 ms | 5× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 29 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 2,90 tok/s TG Prefill 147 · TTFT 70.727 ms | 5× | NVIDIA GeForce RTX 3090 TiAMD Ryzen 9 8945HX with Radeon Graphics · 92 GB RAM | llama.cppwebsocketQ4_K_M | Details → | |
| 30 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 2,88 tok/s TG Prefill 147 · TTFT 70.636 ms | 10× | NVIDIA GeForce RTX 3090 TiAMD Ryzen 9 8945HX with Radeon Graphics · 92 GB RAM | llama.cppwebsocketQ4_K_M | Details → | |
| 31 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 2,05 tok/s TG Prefill 122 · TTFT 82.163 ms | 10× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 32 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,99 tok/s TG Prefill 133 · TTFT 80.073 ms | 5× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 33 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,73 tok/s TG Prefill 95 · TTFT 22.362 ms | 1× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 34 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,39 tok/s TG Prefill 10 · TTFT 211.800 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 35 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,39 tok/s TG Prefill 10 · TTFT 217.448 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 36 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,39 tok/s TG Prefill 10 · TTFT 209.035 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 37 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,38 tok/s TG Prefill 10 · TTFT 211.403 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 38 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,38 tok/s TG Prefill 10 · TTFT 216.229 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 39 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,37 tok/s TG Prefill 10 · TTFT 218.406 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 40 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,37 tok/s TG Prefill 10 · TTFT 218.058 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 41 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,36 tok/s TG Prefill 10 · TTFT 209.786 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 42 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,34 tok/s TG Prefill 10 · TTFT 212.473 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 43 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,33 tok/s TG Prefill 10 · TTFT 213.957 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 44 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,33 tok/s TG Prefill 10 · TTFT 211.672 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 45 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,32 tok/s TG Prefill 10 · TTFT 218.758 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 46 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,90 tok/s TG Prefill 95 · TTFT 22.429 ms | 1× | NVIDIA GeForce RTX 3090 TiAMD Ryzen 9 8945HX with Radeon Graphics · 92 GB RAM | llama.cppwebsocketQ4_K_M | Details → | |
| 47 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,58 tok/s TG Prefill 84 · TTFT 25.390 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliQ4_K_M | Details → | |
| 48 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,01 tok/s TG Prefill 11 · TTFT 423.792 ms | 10× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 49 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,01 tok/s TG Prefill 11 · TTFT 425.889 ms | 5× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 50 | Devstral-2-123B-Instruct-2512123BMistral AI Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,01 tok/s TG Prefill 17 · TTFT 398.361 ms | 5× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawQ4_K_M | Details → |