Decision tools // transparent assumptions
Work out what fits, what it costs, and what to watch.
Start with model memory, then evaluate complete ownership economics. Every price and rate is either a dated reference or an input you control—never an unlabeled live-market claim.
LLM VRAM calculator
Turn a model and quantization into a memory target, add operating headroom, and see which indexed devices clear it.
Model → memory → hardware →GPU cost calculator
Estimate electricity, net ownership cost, effective hourly cost, and an optional buy-versus-cloud crossover.
Hardware + energy → TCO →GPU watch planner
Build a workload-first watchlist from the dated GPU index while automated offer feeds remain disabled and unconnected.
Reference-only beta →