Llama 4 Scout is Meta's first MoE Llama — 109 B total parameters, 17 B per-token compute, 16 experts. The headline feature is a 10M-token context window (per Meta's April 2025 announcement), enabling whole-codebase or whole-book reasoning. Hardware-wise it's awkward: you must hold all 109 B weights in VRAM (~240 GB at FP16, ~66 GB at Q4_K_M from the table below), but per-token compute is only 17 B — so a single H100 80 GB at Q4 is the practical entry point for inference, or 2x H100 for FP16. See our [Q4_K_M quantization](/glossary/q4-k-m) and [MoE architecture](/glossary/moe) glossary entries for the math, and the [/benchmarks](/benchmarks) page for measured tok/s.
Llama 4 Scout 17B needs at minimum 65.9 GB of VRAM at Q4_K_M quantization (239.8 GB at FP16). The cheapest GPU that comfortably fits with KV-cache headroom is the Intel Data Center GPU Max 1550 (128 GB VRAM, $2,500 MSRP). Community benchmark submissions are open. This model exceeds 48 GB at Q4, so plan for a 2-or-more-GPU split.
Weights only. Add ~20-30% for KV cache at typical context lengths.
| Quant | VRAM | Quality | Notes |
|---|---|---|---|
| FP16 | 239.8 GB | Reference | Training-precision weights |
| Q8_0 | 119.9 GB | Near-lossless | 8-bit, ~0.1% perplexity hit |
| Q6_K | 97.1 GB | Very high | 6-bit, near Q8 quality |
| Q5_K_M | 82.7 GB | High | Strong middle ground |
| Q4_K_M | 65.9 GB | Balanced (recommended) | Default for local deployments |
| Q4_0 | 67.4 GB | Legacy 4-bit | Older GGUF, kept for compatibility |
| Q3_K_M | 51.6 GB | Lossy | When VRAM is very tight |
| Q2_K | 38.4 GB | Extreme | Rescue option, quality degrades visibly |
Filtered from MyAIHardware's GPU database. FP16 needs 239.8 GB; Q4_K_M needs 65.9 GB.
| GPU | Per-card VRAM | 2x | 4x | 8x |
|---|---|---|---|---|
| Apple Mac mini (M4 Pro, 64 GB) | 64 GB | 128 | 256 | 512 |
| Apple Mac Studio (M1 Max, 64 GB) | 64 GB | 128 | 256 | 512 |
| Apple Mac Studio (M1 Ultra, 64 GB) | 64 GB | 128 | 256 | 512 |
| Apple Mac Studio (M2 Max, 64 GB) | 64 GB | 128 | 256 | 512 |
| Apple Mac Studio (M2 Ultra, 64 GB) | 64 GB | 128 | 256 | 512 |
| Apple Mac Studio (M4 Max, 64 GB) | 64 GB | 128 | 256 | 512 |
We don't have community benchmarks for this exact model yet.
Official weights + popular runtime tags.
ollama pull llama4:scoutStep-by-step guides that use this model.
Models in a similar size or capability class.
Use these in papers, blog posts, or internal docs.
@misc{llama-4-scout-2025,
title={Llama 4 Scout 17B},
author={Meta AI},
year={2025},
url={https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct}
}Meta AI (2025). Llama 4 Scout 17B [Model card]. Hugging Face. https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct
Llama 4 Scout 17B (Llama 4, Meta AI, 2025). Available at https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct.