Inference · compare

Gemma 3 270M vs Qwen3-VL-8B-Instruct-abliterated-v1.

Side by side only on machines of the same kind, so the difference is the model, not the computer. Each figure is the fastest build of that model on that machine.

1machine kind both ran on
MachineGemma 3 270M tok/sQwen3-VL-8B-Instruct-abliterated-v1 tok/sContext A / B
Windows, 8 CPUs · 16 GB52Q4_K_M (4-bit) by unsloth2.6Q2_K (2-bit) by mradermacher— / 4K