Which model runs on my machine?

Pick your configuration. For each: the open-weight models that fit in memory, the quantization level, the remaining margin and the command to run them. Computed from actual weight sizes, not from promises.