[bmdpat]

Tools

Pick what fits your GPU. Keep the run under control.

10 live / 2 beta from 12 public tools. Start with VRAM, model fit, or quantization. Use AgentGuard when a run needs hard limits.

Observed package downloads, updated hourly. SDK totals use Pepy first, then Pypistats no-mirror overall data when Pepy is unavailable; SDK recent-month counts use Pypistats; MCP recent-month counts use npm. Downloads are a usage signal, not an adoption proof. Tracked copy is shown when metric upstreams are unavailable.

10 live / 2 beta from 12 public tools.

§ 001 / RACK // one surface

Search it, filter it, or sort by how finished it is.

The counts on each chip come from the catalog itself, so a filter can never claim more tools than the rack can show.

Rack12

§ 002 / WHERE TO START

If you only open one of them.

Which tool should I open first?

Start at the sizing desk. It answers the question the rest of the rack depends on: what your card can actually hold, at which quantization, with how much context.

vramquantnum_ctx

Rejected

  • AgentGuardonly once code runs
  • Quant compareafter a model is chosen
  • Dota statsunrelated to local AI
High confidenceOpen the desk

Get the local AI lab notes

New benchmark rows, VRAM fit checks, and model-fit notes from measured runs on owned hardware. One evidence-backed note on Friday when there is something worth sharing.

Get the requested artifact now, then at most one evidence-backed Local AI Lab Note on Friday when there is something worth sharing. One-click unsubscribe. No sponsored placements. Privacy.