research-gguf

A guide to GGUF, a model-file format used by llama.cpp to run language models on CPUs, Apple Silicon, and graphics processors.

In plain words
What is it for?
Use it to select 2- to 8-bit quantization options and deploy models with tools such as LM Studio, Ollama, or llama.cpp.
Why use it?
It helps choose and use reduced-precision model files when running AI locally on consumer hardware.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/graycodeai/starling/research-gguf
Any agent
npx skills add GrayCodeAI/starling --skill research-gguf
Clone the repo
git clone --depth 1 https://github.com/GrayCodeAI/starling

Made for: Claude Code, Codex.

Per session 46 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,128 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin 92% copy Near-identical to another mod in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00046 $0.03128
Opus 5 $0.00023 $0.01564
Sonnet 5 $0.00009 $0.00626
Haiku 4.5 $0.00005 $0.00313

Measured 2d ago against content hash 124319a6f355, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

research-gguf scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

This is a copy

92% identical to gguf-quantization — 13 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

categories/ai-ml/research-gguf/SKILL.md · 427 lines

How it starts

The opening of the file, as written. The whole thing — 427 lines — stays where its author put it; the contents beside it link to each section on GitHub.

GGUF - Quantization Format for llama.cpp

The GGUF (GPT-Generated Unified Format) is the standard file format for llama.cpp, enabling efficient inference on CPUs, Apple Silicon, and GPUs with flexible quantization options.

When to use GGUF

Use GGUF when:

  • Deploying on consumer hardware (laptops, desktops)
  • Running on Apple Silicon (M1/M2/M3) with Metal acceleration
  • Need CPU inference without GPU requirements
  • Want flexible quantization (Q2_K to Q8_0)
  • Using local AI tools (LM Studio, Ollama, text-generation-webui)

Key advantages:

  • Universal hardware: CPU, Apple Silicon, NVIDIA, AMD support
  • No Python runtime: Pure C/C++ inference
  • Flexible quantization: 2-8 bit with various methods (K-quants)
  • Ecosystem support: LM Studio, Ollama, koboldcpp, and more
  • imatrix: Importance matrix for better low-bit quality

Use alternatives instead:

  • AWQ/GPTQ: Maximum accuracy with calibration on NVIDIA GPUs
  • HQQ: Fast calibration-free quantization for HuggingFace
  • bitsandbytes: Simple integration with transformers library
  • TensorRT-LLM: Production NVIDIA deployment with maximum speed

Quick start

Installation

# Clone llama.cpp
git clone https://github.com/ggml-org/llama.cpp
cd llama.cpp

# Build (CPU)
make

# Build with CUDA (NVIDIA)
make GGML_CUDA=1

# Build with Metal (Apple Silicon)
make GGML_METAL=1

# Install Python bindings (optional)
pip install llama-cpp-python

Convert model to GGUF

# Install requirements
pip install -r requirements.txt

# Convert HuggingFace model to GGUF (FP16)
python convert_hf_to_gguf.py ./path/to/model --outfile model-f16.gguf

# Or specify output type
python convert_hf_to_gguf.py ./path/to/model \
    --outfile model-f16.gguf \
    --outtype f16

Quantize model

# Basic quantization to Q4_K_M
./llama-quantize model-f16.gguf model-q4_k_m.gguf Q4_K_M

# Quantize with importance matrix (better quality)
./llama-imatrix -m model-f16.gguf -f calibration.txt -o model.imatrix
./llama-quantize --imatrix model.imatrix model-f16.gguf model-q4_k_m.gguf Q4_K_M

Read the full file on GitHub · 427 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 427 lines · 46 tokens per session scan A 124319a6f355

Subscribe to this mod's changes

research-gguf is a skill published in the GitHub repository GrayCodeAI/starling (2 stars, last pushed 3d ago), licensed MIT. It adds 46 tokens to every session and 3,128 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. It is 92% identical to gguf-quantization, differing in 13 lines, and is treated as a copy.

Related

Other skills, from other repositories

gguf-quantization

GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.

Orchestra-Research/AI-Research-SKILLs · 48 tokens

llama-cpp

Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.

Orchestra-Research/AI-Research-SKILLs · 76 tokens

gguf-quantization

GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.

ihatesea69/HieuNghi-AI-Skills · 48 tokens

llama-cpp

Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.

ihatesea69/HieuNghi-AI-Skills · 76 tokens

docker

Manage Docker containers, images, volumes, and Compose stacks via the docker CLI — list, inspect, logs, run, build, stop, remove, compose up/down. Use for local container ops.

AtomicBot-ai/atomic-agent · 44 tokens

gog-workspace

Use the gog CLI for Google Workspace tasks across Gmail, Calendar, Drive, Docs, Sheets, Contacts, and related services. Use when the user asks to check email, search Gmail, inspect calendar events, find Drive files, read Docs or Sheets, or manage Google Workspace data through gog.

AtomicBot-ai/atomic-agent · 68 tokens