LLM VRAM calculator

Work out how much GPU memory a large language model needs for weights, KV cache and activations — and which GPUs can host it.

Quick preset

Parameter count (B)

Weights precision

Context length (tokens)

Batch size

Number of layers

KV heads

Head dimension

KV cache is computed in FP16 (2 bytes) for {2 × layers × kv_heads × head_dim × context × batch}. Activations and framework overhead are estimated at 15% of weight size.

Frequently asked questions

How accurate is the GPU cost estimator?

It multiplies the hourly rate we collected by your expected running hours and adds storage and egress at the stated assumptions. It is a budgeting tool, not a quote: real invoices include taxes, idle time and any minimum billing increments the provider applies.

How does the LLM VRAM calculator work?

It computes weight memory from parameter count and precision, KV cache from the attention geometry (2 x layers x KV heads x head dimension x context x batch, in FP16), and adds 15 percent of the weights for activations and framework overhead.

Why does my measured VRAM usage differ from the estimate?

Real usage depends on the serving framework, whether KV cache is paged or preallocated, CUDA graph capture, and how aggressively the engine reserves memory. Treat the number as a sizing guide and leave headroom.

Can I share a configuration?

Yes. The share button copies a link containing your inputs in the URL fragment (after the # symbol). The fragment never leaves the browser, so the shared link reveals nothing to our servers.