GHOSTAI — Christmas Spirit GGUF (7B)
A holiday-forward 7B model tuned for cozy storytelling, cheerful roleplay, and warm seasonal vibes — shipped in GGUF for the llama.cpp ecosystem. Quantized builds included for most users. An optional F16 GGUF is included for maximum fidelity.
Overview
GHOSTAI: Christmas Spirit is designed to produce cozy, wholesome, and festive outputs: winter scenes, gift-giving stories, cheerful dialog, holiday recipes, and family-friendly roleplay.
This repository provides multiple GGUF variants, so you can choose the best balance of quality, speed, and memory usage for your hardware.
You can run it:
CPU-only
With GPU offload (CUDA / Metal / Vulkan builds of llama.cpp)
Quant choice is independent of CPU vs GPU; GPU offload is controlled by runtime flags (example: -ngl).
Files (this release)
Sizes below reflect the exported files in this repo.
File
Quant
Approx size
Rough RAM needed (4k ctx)
christmas_mistral_v1.f16.gguf
f16
~13.5 GB
~16–18 GB
christmas_mistral_v1.Q8_0.gguf
Q8_0
~7.2 GB
~10–11 GB
christmas_mistral_v1.Q6_K.gguf
Q6_K
~5.5 GB
~8–9 GB
christmas_mistral_v1.Q5_K_M.gguf
Q5_K_M
~4.8 GB
~7–8 GB
christmas_mistral_v1.Q5_K_S.gguf
Q5_K_S
~4.7 GB
~7–8 GB
christmas_mistral_v1.Q4_K_M.gguf
Q4_K_M
~4.1 GB
~6–7 GB
christmas_mistral_v1.Q4_K_S.gguf
Q4_K_S
~3.9 GB
~6–7 GB
christmas_mistral_v1.Q3_K_M.gguf
Q3_K_M
~3.3 GB
~5–6 GB
christmas_mistral_v1.Q3_K_S.gguf
Q3_K_S
~3.0 GB
~5–6 GB
christmas_mistral_v1.Q2_K.gguf
Q2_K
~2.5 GB
~4–5 GB
christmas_mistral_v1.TQ1_0.gguf
TQ1_0
~1.6 GB
~3–4 GB
RAM notes (rough):
Assumes ~4k context and typical llama.cpp overhead.
For 8k context, plan +1–2 GB extra (or more depending on runner/settings).
GPU offload can shift some load to VRAM; you still need system RAM.
1./llama-cli \2 -m christmas_mistral_v1.Q4_K_M.gguf \3 -ngl 0\4 -c 4096\5 -p "You are GHOSTAI Christmas Spirit. Write a cozy winter story set in a small town, with a warm ending."
