Meta
Llama 4 — all variants, VRAM & hardware
Meta's Llama 4 lineup — Scout 109B MoE and Maverick 400B MoE with up to 10M context. Strong multilingual and reasoning performance, permissive community license.
Variant comparison
| Variant | Params | Active | VRAM @ Q4 | Architecture | |
|---|---|---|---|---|---|
| Llama 4 Scout 17B 16E | 109B | 17B | ~66.5 GB | moe | Details → |
| Llama 4 Maverick 17B 128E | 400B | 17B | ~244.0 GB | moe | Details → |
Deep-dive guides
Llama 4 VRAM & hardware guides
FAQ
Frequently asked questions
What is Llama 4?
Llama 4 is Meta's meta's llama 4 lineup — scout 109b moe and maverick 400b moe with up to 10m context. strong multilingual and reasoning performance, permissive community license.
How many variants are in the Llama 4 family?
We track 2 canonical Llama 4 variants in our catalog, spanning 109B to 400B parameters.
Which Llama 4 variant should I run?
Pick the largest variant that fits on your hardware at Q4_K_M quantization. For 8 GB GPUs, the smallest dense variants. For 24 GB GPUs, the 27B dense or 35B MoE (if available). For workstation / Mac Studio, the 122B or larger MoE. Use our fit calculator on each model page for the exact verdict.
See also