GGUF Quantization Deep Dive: Q2 vs Q4 vs Q8 Explained

GGUF quantization reduces model weights from 16-bit floats to 2-8 bits per weight. The difference between Q2 and Q8 can mean 4x memory savings or near-lossless quality — knowing which to pick is one of the most impactful decisions in local LLM setup.

Quantization Types Explained

FormatBits/WeightQuality vs FP16Size (8B model)Use When
Q2_K2.6 bits-15 to -25%2.7 GBOnly when VRAM is extremely tight
Q3_K_M3.3 bits-8 to -12%3.4 GBMinimum viable for chat
Q4_K_M4.8 bits-3 to -5%4.8 GBBest everyday balance
Q5_K_M5.7 bits-1 to -2%5.7 GBQuality-first with VRAM headroom ★
Q6_K6.6 bits-0.5 to -1%6.6 GBNear-lossless with 24GB+ VRAM
Q8_08.5 bits~0%8.5 GBEffectively lossless
F1616 bitsBaseline16 GBFine-tuning, reference only

The K-Quant Suffix Explained

Suffixes like _K_M, _K_S, _K_L refer to how quantization is applied to different weight matrices:

  • K: Uses K-quants algorithm (better quality than non-K at same bit count)
  • _S / _M / _L: Small/Medium/Large — affects which layers get higher precision. _M is the best default.
  • _0: Simple block quantization (older format, less efficient)

Recommended Choices by VRAM

VRAM AvailableBest ChoiceMax Model Size
8 GBQ5_K_M for 7B / Q4_K_M for 8B~8B at Q4
12 GBQ5_K_M for 8B / Q4_K_M for 14B~14B at Q4
16 GBQ5_K_M for 14B / Q4_K_M for 32B (partial)~32B at Q3
24 GBQ5_K_M for 32B~32B fully on GPU

Practical Example: Downloading Specific Quants

# Pick Q5_K_M for best quality when VRAM allows
huggingface-cli download bartowski/Qwen2.5-32B-Instruct-GGUF   --include "*Q5_K_M*"   --local-dir ./models/qwen-32b-q5

# Q4_K_M for tighter VRAM budget
huggingface-cli download bartowski/Qwen2.5-32B-Instruct-GGUF   --include "*Q4_K_M*"   --local-dir ./models/qwen-32b-q4

For which GPUs have enough VRAM for different quant levels, see LLMHardware.io's VRAM calculator.

Comments

Post a Comment

Popular posts from this blog

AMD RX 9070 XT for Local LLMs: ROCm Performance Guide (2026)

The AMD RX 9070 XT launched in early 2026 at $599 with 16GB GDDR6 and 648 GB/s memory bandwidth. For local LLM users who want a mid-range GPU without NVIDIA prices, this changes the calculus significantly. Here is what it actually delivers. RX 9070 XT Specs vs Competition GPU VRAM Bandwidth Price LLM Score RX 9070 XT 16 GB 648 GB/s $599 ~60 tok/s RTX 4070 Ti Super 16 GB 672 GB/s $799 ~62 tok/s RTX 4080 Super 16 GB 736 GB/s $999 ~68 tok/s RTX 4090 24 GB 1,008 GB/s ~$1,100 ~90 tok/s At $599 with 16GB VRAM, the RX 9070 XT is the best value GPU for local LLMs in the mid-range tier in 2026. See the full comparison at LLMHardware.io's AMD GPU guide . ROCm Setup on Linux (2026) # Ubuntu 22.04/24.04 sudo apt install rocm-hip-libraries hip-base # or use the AMD installer: wget https://repo.radeon.com/amdgpu-install/6.3/ubuntu/jammy/amdgpu-install_6.3.60300-1_all.deb sudo dpkg -i amdgpu-install*.deb sudo amdgpu-install --usecase=rocm,lrt,openclsdk # Verify ROCm sees your GPU r...
Read more

llama.cpp flash attention: how to enable and benchmark

Image
Flash attention computes attention with far less memory traffic. The math is identical; the memory movement is not. In llama.cpp it matters most in two situations: long context windows and tight video memory (VRAM). With flash attention enabled, the key-value cache shrinks substantially. A 32k-token context window becomes realistic on a 16GB card, and the prompt-processing phase speeds up, so long documents get ingested faster. The catch is compatibility: not every model and quantization combination supports it, and on some setups the speedup is modest. If you run short chats on a strong GPU, you may barely notice. Where it shines is retrieval-augmented generation (RAG) and long-document work on mid-range cards. Enable it, compare your tokens per second, and keep it if your workload benefits. For the common case of pasting a long PDF and asking questions, it is close to free performance. See how context length changes VRAM needs .
Read more

SillyTavern with Local LLMs: Complete Setup for AI Companions

Image
SillyTavern is the most feature-rich frontend for character-based AI conversations. What began as an AI roleplay tool has evolved into a powerful platform with persona management, memory systems, extensive prompt engineering tools, and connections to every major local inference backend. Here's how to set it up with local models. Quick connect: SillyTavern connects to Ollama, KoboldCpp, llama.cpp server, and TabbyAPI (ExLlamaV2) out of the box. Any of the hardware setups at LLMHardware.io provides a solid backend for SillyTavern. Installation Install via git: git clone https://github.com/SillyTavern/SillyTavern && cd SillyTavern && npm install then node server.js . Access at localhost:8000. For persistent installation, use the included start.sh script. SillyTavern uses Node.js and runs on any platform. The companion app (STMP) enables mobile access to your SillyTavern instance via Tailscale or local network. Character Cards and the TavernAI Format Chara...
Read more