🏆
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 | Qwen2.5-3B-Instruct3BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 259,86 tok/s TG Prefill 9.253 · TTFT 3.361 ms | 10× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz · 504 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 2 | Qwen2.5-3B-Instruct3BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 225,82 tok/s TG Prefill 9.739 · TTFT 3.277 ms | 10× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz · 504 GB RAM | llama.cppgodclawQ8_0 | Details → | |
| 3 | Qwen2.5-3B-Instruct3BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 187,50 tok/s TG Prefill 6.698 · TTFT 2.090 ms | 5× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz · 504 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 4 | Qwen2.5-3B-Instruct3BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 168,19 tok/s TG Prefill 7.879 · TTFT 1.850 ms | 5× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz · 504 GB RAM | llama.cppgodclawQ8_0 | Details → | |
| 5 | Qwen2.5-3B-Instruct3BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 106,40 tok/s TG Prefill 3.776 · TTFT 651 ms | 1× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz · 504 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 6 | Qwen2.5-3B-Instruct3BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 73,93 tok/s TG Prefill 4.119 · TTFT 596 ms | 1× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz · 504 GB RAM | llama.cppgodclawQ8_0 | Details → |