AI model compatibility index
Find the best model your machine can run.
Choose your hardware and workload. We’ll rank compatible Ollama tags and show exactly why each one made the cut.
Recommendations
Loading model catalog…
No compatible models found
Try a smaller context, enable partial offload, or choose a GPU with more memory.
FitFull GPU, partial offload, or unavailable for your memory budget.
SpeedDecode tokens per second from a hardware model; only L4 coefficients are measured.
CapabilityIntelligence from stored benches × quant when available; otherwise a size prior with exponential age decay (half-life 1.1y).
Task fitUse-case score from a task-specific bench mix (same shape as intelligence); size prior when that mix is missing.
VRAMWeights + projector + capped KV + graph memory at your selected context.
DataSourced from the Ollama library and model registries; rebuilt from prod/data on GitHub.