LocalClaw/LLM/Llama 4 Maverick (17B/128E MoE)
Open-weight local LLM

Llama 4 Maverick (17B/128E MoE)

Catalogue summary: Meta's largest open MoE. 17B active params across 128 experts (~400B total). Multimodal with exceptional image reasoning. Server-grade hardware required. Llama 4 License.

Repository editorial metadata; verify comparative claims in the linked upstream material.

320 GB catalogue minimum320 GB RAMQ4_K_MVision tasks
Choose an app

This is an advanced workstation/server GGUF path. Unsupported apps are clearly marked.

Compare models
LM StudioNot available for this model
Open in UnslothLaunches Unsloth Desktop on this model's download page
OllamaNot available for this model
Open on Hugging FaceFiles, licence and available downloadsOpen llama.cpp setupAdvanced option: opens GGUF files and patch notesUse with LocalClawOptional workspace after the model is installed

Desktop app links require the app to be installed. If nothing opens, LocalClaw will show app-download and model-file fallbacks.

Parameters
17B active (400B total, 128 experts)
Minimum RAM
320 GB
Model size
220 GB
Quantization
Q4_K_M

Can Llama 4 Maverick (17B/128E MoE) run locally?

Llama 4 Maverick (17B/128E MoE) has a catalogue minimum of 320 GB RAM with Q4_K_M. Actual memory use and speed vary by context length, runtime, backend and system headroom.

Treat llama-4-maverick as an advanced workstation/server-grade catalogue target with verified GGUF artefacts. Open the GGUF files and follow the upstream runtime notes; LocalClaw does not claim a one-click LM Studio install.

Model sourcePublic GGUF repository (verified 2026-09-23): unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUFUpstream release
chatvisionquality

Deployment path

01
Check RAM fitServer-grade target. Plan for 320 GB class multi-GPU memory.
02
Load the modelOpen the verified GGUF files for llama-4-maverick and follow the upstream runtime or patch notes first.
03
Control locallyUse LocalClaw to manage models, agents, chat, channels and scheduled OpenClaw work.

Catalogue record

  • Family: llama
  • Parameters: 17B active (400B total, 128 experts)
  • Recommended quantization: Q4_K_M
  • Catalogue minimum RAM: 320 GB
  • Catalogue model size: 220 GB
  • Tags: chat, vision, quality

Practical limits

  • Catalogue RAM is a minimum estimate, not a guarantee for every context length or runtime.
  • Speed and memory use vary by quantization, backend, context length and system headroom.
  • Verify architecture, licence and usage restrictions in the linked upstream material before deployment.

Catalogue tags

  • chat
  • vision
  • quality

Capability profile

Repository catalogue ratings used by LocalClaw's editorial rubric. They are not a standardized third-party benchmark.

speed
1
quality
10
coding
10
reasoning
10

Technical notes

Developer
Meta AI
License
See upstream repository
Context window
131,072 tokens
Architecture
Mixture of Experts (MoE) — 400B total with native vision

This model fits these next steps

Hardware fit is based on LocalClaw's RAM tier, model size and quantization metadata. Always leave memory headroom for your OS and runtime.

Very large memoryMac Studio Ultra classCheck model size firstNVIDIA GB10 / server optionsMore practical alternativesCompare smaller models

Related catalogue entries

Linked mechanically by family, shared tags and nearby RAM tier; this is not a quality ranking.

Llama 4 Maverick (17B/400B MoE) 400B (17B active, 128 experts)Llama 4 Scout (17B/109B MoE) 109B (17B active, 16 experts)Llama 3.2 Vision (90B) 90BLlama 3.3 (70B) 70BLlama 4 Scout (17B/16E MoE) 17B active (109B total, 16 experts)Llama 3.1 (70B) 70BLlama 3.2 Vision (11B) 11BLlama 3 (8B) 8B

Where to go next