ChrisColeTech/z-image-turbo-GGUF

Z-Image Turbo — GGUF build

6B text-to-image, distilled for a 9-step CFG-free profile. Repacked as a split GGUF layout.

What this repo is: a requantized, split-layout build of the upstream model — weights only, not a retrain. The recommended settings below are the values these weights are actually run with day to day, not copied from the upstream card.

---

Samples

**prompt:** `a red apple on a rustic wooden table beside a window, soft daylight, visible wood grain, studio photograph, sharp focus` — 1024×1024, 9 steps, guidance 0.0, seed 7	**prompt:** `portrait of an older fisherman with a weathered face, natural window light, shallow depth of field, photorealistic` — 1024×1024, 9 steps, guidance 0.0, seed 99
**prompt:** `a rain-slicked Tokyo street at night, neon signs reflecting in puddles, cinematic, 35mm photograph` — 1024×1024, 9 steps, guidance 0.0, seed 1234

---

img2img

This model does img2img as well as txt2img. The left image is the input; the right is what this model made of it.

**source prompt:** `a red apple on a rustic wooden table beside a window, soft daylight, visible wood grain, studio photograph, sharp focus` — 1024×1024, seed 7 — generated separately as the img2img input	**img2img prompt:** `the same scene rendered as a vivid oil painting, thick visible brushstrokes, rich saturated colour, painterly texture` — 1024×1024, **9 steps**, **strength 0.9**, seed 7 — about 6 s on an A100 80GB

strength has a hard threshold on this model. A sweep at 0.3 / 0.5 / 0.7 / 0.9 — same source, prompt and seed — produced no meaningful restyle below 0.9; those values add gloss and surface detail and leave the photograph a photograph. The painterly transform appears only at 0.9. Treat it as near-binary rather than a dial: 0.9 to restyle, low values only for a light finishing pass.

Recommended settings

Values this build is run with in practice. They are read out of a live config when this card is generated, so they cannot drift from what is actually used.

Parameter	Production value	Meaning
width	1024	Output width in pixels
height	1024	Output height in pixels
steps	9	Denoising steps
guidance	0.0	Guidance scale
preview_every	3	Emit a TAESD preview every N steps
strength	0.9	img2img denoise strength (1.0 = ignore the input image)

Supported modes: txt2img, img2img

Notes and gotchas

• Guidance is 0.0, not 1.0. Z-Image Turbo is a guidance-distilled turbo model; this build expects guidance=0.0 and any positive CFG fights the distillation.
• Text encoder is Qwen3-VL-4B, not the encoder in the original upstream layout. The transformer file also carries a -fresh suffix; both are reflected in the paths below.
• 9 steps at 1024² is the verified profile for this build.

---

Layout

Components ship as separate files rather than one bundle: the transformer, the text encoder(s) and the VAE each load independently, with configs and tokenizers under config/. Any loader that accepts explicit per-component paths can consume this directly — point it at the files listed below.

A generic diffusers snippet is deliberately omitted. This is a repacked split layout with substituted encoders, so an upstream example will not load it unmodified, and an untested snippet would be worse than none.

---

Files

File	Size	Role
split/z-image-turbo-q6k-fresh.gguf	4.71 GB	transformer (main weights)
split/text_encoders/Qwen3-VL-4B-Q4_K_M.gguf	2.32 GB	text encoder
split/vae/ae.safetensors	319.77 MB	VAE
config/tokenizer/tokenizer.json	6.71 MB	tokenizer / processor
config/tokenizer/vocab.json	2.65 MB	tokenizer / processor
config/tokenizer/merges.txt	1.59 MB	tokenizer / processor
config/tokenizer/tokenizer_config.json	9.50 KB	tokenizer / processor
config/vae/config.json	805 B	VAE
config/text_encoder/config.json	726 B	text encoder
config/controlnet/config.json	542 B	config
config/transformer/config.json	473 B	transformer (main weights)
config/model_index.json	467 B	config
config/scheduler/scheduler_config.json	173 B	scheduler config

---

Provenance

• Upstream base model: Tongyi-MAI/Z-Image-Turbo (upstream license: apache-2.0)
• This build: requantized and relaid out into a split component layout. Weights are not retrained; only the format and directory layout differ.
• License: left as unknown in this repo's metadata. Refer to the upstream model's license for redistribution and commercial-use terms — several of these bases are non-commercial.