Will it load? 🧮

LM Studio refuses to load a model with “insufficient system resources”, but there seems to be enough RAM? This calculator reproduces LM Studio's model-loading guardrail so you can see the budget, the estimate and the reason. It's unofficial, open source, and runs entirely in your browser.

1 · Your machine

Guardrail budget: –

2 · Already loaded in LM Studio

Models you loaded via lms load or the GUI stay loaded. Just-in-time (API) loads only evict other just-in-time models. Other apps are not counted by the guardrail.

3 · Model to load

How the guardrail decides

  1. Budget: Strict 65 %, Balanced 75 %, Relaxed 85 % of the first 16 GiB of memory, plus 85 / 85 / 90 % of everything above. macOS counts RAM only; Windows/Linux add dedicated VRAM. Custom uses your fixed number.
  2. Estimate: MLX (safetensors) and embedding models are estimated as file size × 1.4, independent of the context length you pick. GGUF LLMs get a detailed estimate from context length, layers and KV cache — check those with lms load <model> --estimate-only.
  3. Decision: it loads if memory used by LM Studio's loaded models + estimate ≤ budget. A loaded model counts as its file size plus about 0.6 GiB; with nothing loaded, nothing counts.
Check your own machine from the terminal 
uvx --from git+https://github.com/nikpottbecker/will-it-load will-it-load
uvx --from git+https://github.com/nikpottbecker/will-it-load will-it-load verify

verify compares every prediction with LM Studio's own dry run (lms load --estimate-only) — no model gets loaded.