🏆
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 | Step-3.5-FlashStepFun Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 16,78 tok/s TG Prefill 165 · TTFT 73.137 ms | 5× | 2x AMD Radeon RX 7900 XTXAMD Ryzen Threadripper PRO 3955WX 16-Cores · 189 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 2 | Step-3.5-FlashStepFun Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 15,35 tok/s TG Prefill 219 · TTFT 122.535 ms | 10× | 2x AMD Radeon RX 7900 XTXAMD Ryzen Threadripper PRO 3955WX 16-Cores · 189 GB RAM | llama.cppgodclawQ4_K_M | Details → | |
| 3 | Step-3.5-FlashStepFun Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 11,46 tok/s TG Prefill 99 · TTFT 26.457 ms | 1× | 2x AMD Radeon RX 7900 XTXAMD Ryzen Threadripper PRO 3955WX 16-Cores · 189 GB RAM | llama.cppgodclawQ4_K_M | Details → |