Avg Speed44.59 t/s
Latency0.87 s
Total Tests5
Models1
Updated3/16/2026
Created At3/15/2026
A
llama.cpp
llama.cpp is an open-source inference engine for running large language models locally, optimized for CPU and GPU performance.
Community
Join LMSpeed QQ Group
Join us for speed-test updates and support. Group ID:
llama.cpp is an open-source inference engine for running large language models locally, optimized for CPU and GPU performance.
| Model | Speed | Latency | Tests |
|---|---|---|---|
| Qwen3.5-35B-A3B-Q4_K_M.gguf | 44.59 t/s | 0.87s | 5 |
| Time | Model | Speed | Latency |
|---|---|---|---|
| Mar 15, 01:02 PM | Qwen3.5-35B-A3B-Q4_K_M.gguf | 44.59 t/s | 0.87s |