🏆
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 | Qwen2.5-0.5B-InstructQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 19,64 tok/s TG Prefill 135 · TTFT 103.828 ms | 5× | Keine GPU (CPU-only) i4 x Cortex-A78C + 4 x Cortex-A55 · 15 GB RAM | llama.cppopenclaw_cliQ4_K_M | Details → | |
| 2 | Qwen2.5-0.5B-InstructQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 19,33 tok/s TG Prefill 171 · TTFT 178.576 ms | 10× | Keine GPU (CPU-only) i4 x Cortex-A78C + 4 x Cortex-A55 · 15 GB RAM | llama.cppopenclaw_cliQ4_K_M | Details → | |
| 3 | Qwen2.5-0.5B-InstructQwen (Alibaba) Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 11,74 tok/s TG Prefill 85 · TTFT 29.146 ms | 1× | Keine GPU (CPU-only) i4 x Cortex-A78C + 4 x Cortex-A55 · 15 GB RAM | llama.cppopenclaw_cliQ4_K_M | Details → |