What can you run locally?

LLM VRAM calculator — estimate model memory, or find models for your hardware.

“Will it fit?” is the judgement after model, artifact, workload, runtime, and topology meet — not a third starting point.

Loading catalog snapshot…

Search modelsModelQuantization

Uses a published artifact with a fixed rule. The result always shows what was actually selected.

Hardware

Workload

Context length (cached tokens)

Concurrent sequences

Advanced settings

KV cache typeRecurrent / conv state dtypeEngineDeploymentIntra-replica splitDevice countReplica groupsDevices per replicaOverhead scenarioSystem / display reserve

GiB (2^30 bytes)

Budget already excludes system reserveExtra safety marginCustom bits per weight Include heavier prefill workspace Include theoretical estimates (expert) Allow a substitute if the filter has no matchDisplay unit

Custom model structure

Import local config.json

Parsed in the browser. The file is not uploaded.

Public Hugging Face repoRevision

Only huggingface.co, config.json, no token. Private repos are rejected.

Total parametersActive parametersLayersQuery headsKV headsHead dimension

Estimate

Search or pick a catalog checkpoint to see memory demand.

Search or pick a catalog checkpoint to see memory demand.Result summary