⚡
Performance benchmark
How fast is the model?
Raw throughput on real hardware – how many tokens a model generates per second, how quickly it processes the prompt and how short the time to first token is. Higher is better.
⚡ Generation (tok/s)🔄 Prefill⏱️ TTFT👥 Concurrency
| # | Model / Maker | Metrics | Parallel | GPU / CPU / RAM | Runtime | ||
|---|---|---|---|---|---|---|---|
| 1 | Qwen3-0.6B0.6BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 302,60 tok/s TG Prefill 4.126 · TTFT 3.252 ms | 5× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 2 | Qwen3-0.6B0.6BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 252,73 tok/s TG Prefill 1.840 · TTFT 1.356 ms | 1× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 3 | Qwen3-1.7B1.7BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 202,40 tok/s TG Prefill 2.042 · TTFT 6.616 ms | 5× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 4 | Qwen3-1.7B1.7BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 165,64 tok/s TG Prefill 1.140 · TTFT 2.146 ms | 1× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 5 | Qwen3-4B4BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 101,48 tok/s TG Prefill 945 · TTFT 14.455 ms | 5× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 6 | Qwen3-4B4BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 92,71 tok/s TG Prefill 620 · TTFT 3.961 ms | 1× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 7 | Qwen3-8B8.2BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 68,07 tok/s TG Prefill 577 · TTFT 23.820 ms | 5× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 8 | Qwen3-8B8.2BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 58,58 tok/s TG Prefill 373 · TTFT 6.570 ms | 1× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 9 | Qwen3-14B14BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 10,58 tok/s TG Prefill 185 · TTFT 13.217 ms | 1× | AMD Oberon (PlayStation 5)AMD Zen2 8-Core (PlayStation 5) · 16 GB RAM | llama.cppgodclawQ4_K_M | Details → |