Is 16GB VRAM enough for Llama 4 Scout?

Llama 4 has mixed chunked and full attention. The calculator estimates its memory components; artifact and runtime allocations still determine actual peak memory.

Can I run Llama 4 Maverick on RTX 5090?

The calculator estimates Llama 4 cache components. Multi-GPU and partial offload are outside the current profile; memory capacity alone cannot certify execution on an RTX 5090.

What is the difference between GGUF, EXL2, and AWQ?

GGUF metadata and tensor tables can be inspected. EXL2 and AWQ require different runtime support and are outside this llama.cpp profile.

How much VRAM do I need to run a local LLM?

Memory depends on the artifact, architecture, context and runtime. Current results are uncalibrated estimates and do not establish whether a configuration will run.

How much VRAM does Llama 4 Scout Q4_K_M use?

Select a GGUF artifact to inspect its tensor payload, or choose an assumed BPW conversion scenario. Llama 4 cache follows its mixed layer structure; the total remains an uncalibrated planning estimate.

Why is my local LLM generating text so slowly?

Generation speed requires runtime measurements. This calculator does not diagnose speed or calculate a partial-offload plan.

LLM Memory Calculator · HF and GGUF

Search by model name

Popular models first. Scroll or type a model name.

Select LLM Model

Checking for current models…

Weight precision / conversion scenario

Context Window

Tokens per sequence, including prompt and generated tokens. Cache allocation depends on the architecture.

Your GPU

Available GPU memory (GiB)

KV Cache Precision

Includes quantization block scales. Execution support depends on your llama.cpp build and device.

GPU Count

Multi-GPU allocation and compatibility are pending validation.

Concurrent Requests

Context above is per sequence. This profile uses a unified cache; recurrent state is counted per sequence.

Extra runtime budget (GiB)

An editable planning allowance for compute buffers and other runtime allocations. The initial 1 GiB is not a measured prediction or guaranteed margin.

Prompt micro-batch (llama.cpp -ub)

Affects cache and compute buffers. Effective size is capped by the prompt batch limit and context. Flash Attention is assumed enabled.

Prompt batch limit (llama.cpp -b)

Tokens per logical prompt batch. This is separate from concurrent requests.

Recurrent rollback snapshots per sequence

For recurrent/hybrid architectures. Match the runtime setting; speculative decoding can require additional state.

Remember my last setup on this device

Optional local storage only. Shared links take priority. Uncheck to forget the saved setup.

ℹ

Choose a model to get started

0.00 GiB

Memory planning estimate

Listed capacity: —

Hardware Guide & FAQ

Llama 4 uses mixed chunked and full attention. This calculator reads that structure and estimates its cache. Actual peak memory also depends on the artifact and runtime; 16 GiB capacity alone does not establish compatibility.

GGUF is inspected through its metadata and tensor tables. EXL2 and AWQ require different execution support and are outside this llama.cpp profile. A numeric BPW scenario does not validate those artifacts.

Generation speed depends on the runtime and workload. This calculator does not measure speed or determine how much memory to offload to system RAM.

This calculator does not certify hardware purchases. Memory capacity alone cannot establish whether a model will run under a particular runtime configuration.

Longer context increases full-attention cache. Sliding layers and recurrent state use different allocation rules. Compute buffers and actual peak GPU memory remain uncalibrated; contexts beyond the declared model limit are rejected.

Local LLM VRAM Calculator

Hardware Guide & FAQ

Popular VRAM Calculations

Hardware Guide & FAQ

Found this useful? Share it with someone who needs it.

Popular VRAM Calculations

☁️ How to run ANY model in 3 minutes

Deploy WebUI Template

Download a GGUF Model

Load & Don't Forget to Stop!

You deserve the most accurate tool