Don’t know which model to look up? Start from your graphics card instead: see what LLM your GPU can run. Check my GPU →
How this result is calculated
Matching GPU options
smallest viable pools firstCalculator assumptions
Weights: parameters x bits per weight / 8.
KV cache: layers x KV heads x head dimension x context x batch x K/V tensors x bytes per element. Architecture-aware: MLA and sliding-window models cache far less. Quantizing the cache with -ctk/-ctv shrinks it further.
Runtime: backend overhead covers allocator reserve, graph buffers, CUDA/runtime workspace, and practical safety margin.
Hardware: multi-GPU fits include extra headroom for PCIe and tensor-parallel synchronization.
GGUF files: read entirely in your browser, header only. The file is never uploaded. Weight size is measured from the file itself.
Domain for sale: voxllm.com
Got your number? Now see it the other way round: what your GPU can run →
Architectural Infrastructure & Hardware Math
Deploying open-source models requires absolute precision, not guesswork. This tool serves as the definitive gguf vram calculator, allowing engineers to properly architect their local llm hardware stack before provisioning any physical components. By calculating exact parameter weights, KV cache at various context lengths, and 15% system runtime overhead, this gpu memory calculator ensures your deployment fits flawlessly within your available silicon architecture.