[bmdpat]

§ 001 / SIZING DESK

See what your GPU can actually run.

Pick the GPU you own. Get the local models that fit it, the quant to use, expected tokens per second, and a copy-paste run command. Measured on real hardware where we have it, estimated everywhere else.

Workload
Optimize for
Context length

Pick the GPU you own to see the local models it can run.

What Can My GPU Run? Local LLM Sizing Desk (2026) | Patrick Hughes