Inference · hardware

What runs on Linux PC with an NVIDIA GPU, 32 GB.

Every open model a machine of this kind has run on the network, at its fastest build, fastest first. Speed is tokens a second, measured and signed by the machine itself.

2models measured
239fastest tok/s
2machines
ModelFastest build heretok/sContext$ / M tokens
Qwen2.5 0.5B InstructQ4_K_M (4-bit) by Qwen23932K$0.13
Gemma 4 E2BQ4_K_M (4-bit) by unsloth78128K$0.43