🏆
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 | Qwen3.8-Flash-Next180BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 71,73 tok/s TG Prefill 560 · TTFT 40.628 ms | 10× | 2x NVIDIA RTX A6000AMD EPYC 7203P 8-Core Processor · 252 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 2 | Qwen3.8-Flash-Next180BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 60,89 tok/s TG Prefill 409 · TTFT 24.788 ms | 5× | 2x NVIDIA RTX A6000AMD EPYC 7203P 8-Core Processor · 252 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 3 | Qwen3.8-Flash-Next180BQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 36,99 tok/s TG Prefill 221 · TTFT 12.961 ms | 1× | 2x NVIDIA RTX A6000AMD EPYC 7203P 8-Core Processor · 252 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → |