Inference · compare

Gemma 3 270M vs Qwen2.5 0.5B Instruct.

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.

3machine kinds both ran on
MachineGemma 3 270M tok/sQwen2.5 0.5B Instruct tok/sContext A / B
Linux, 4 CPUs · 31 GB37Q4_K_M (4-bit) by unsloth239Q4_K_M (4-bit) by Qwen— / 32K
Intel(R) Iris(R) Xe Graphics · 1 GB69IQ2_M (2-bit) by unsloth23Q4_K_M (4-bit) by Qwen— / 8K
GeForce RTX 3080 Ti · 12 GB286Q4_K_M (4-bit) by unsloth248Q4_K_M (4-bit) by Qwen— / 8K