Z-image turbo maybe im missing something ?

[chaos4u]

so i have relented and followed the instructions in the release notes using the models linked.

z_image_turbo-Q4_0.gguf

Qwen_3_4b-FP16.gguf

flux2vae.safetensors

and i still can not get this to load,

[Error: KCPP SD Failed to create context!
If using Flux/SD3.5, make sure you have ALL files required (e.g. VAE, T5, Clip...) or baked in!
Otherwise, if you are using GGUF format, you can try the original .safetensors instead (Comfy GGUF not supported)
Load Image Model OK: False]

i have even used the default settings on startup not changing anything, and it seems like i dont have something configured correctly ?
what am i not doing? im under the impression i just needed the 3 files linked , and loaded as per the instructions. however im not able to load z-image turbo despite my efforts.

---

---

Welcome to KoboldCpp - Version 1.103
For command line arguments, please refer to --help

---

Auto Selected CUDA Backend (flag=0)

Loading Chat Completions Adapter: C:\Users\Nitra\AppData\Local\Temp_MEI117922\kcpp_adapters\AutoGuess.json
Chat Completions Adapter Loaded
System: Windows 10.0.19041 AMD64 AMD64 Family 23 Model 1 Stepping 1, AuthenticAMD
Detected Available GPU Memory: 24564 MB
Detected Available RAM: 25477 MB
Initializing dynamic library: koboldcpp_cublas.dll

Namespace(admin=False, admindir='', adminpassword='', analyze='', batchsize=512, benchmark=None, blasthreads=None, chatcompletionsadapter='AutoGuess', cli=False, config=None, contextsize=8192, debugmode=0, defaultgenamt=768, draftamount=8, draftgpulayers=999, draftgpusplit=None, draftmodel=None, embeddingsgpu=False, embeddingsmaxctx=0, embeddingsmodel='', enableguidance=False, exportconfig='', exporttemplate='', failsafe=False, flashattention=False, forceversion=0, foreground=False, genlimit=0, gpulayers=-1, highpriority=False, hordeconfig=None, hordegenlen=0, hordekey='', hordemaxctx=0, hordemodelname='', hordeworkername='', host='', ignoremissing=False, jinja=False, jinja_tools=False, launch=True, lora=None, loramult=1.0, lowvram=False, maingpu=-1, maxrequestsize=32, mmproj=None, mmprojcpu=False, model=[], model_param=None, moecpu=0, moeexperts=-1, multiplayer=False, multiuser=1, noavx2=False, noblas=False, nobostoken=False, nocertify=False, nofastforward=False, nommap=False, nomodel=False, noshift=False, onready='', overridekv=None, overridenativecontext=0, overridetensors=None, password=None, port=5001, port_param=5001, preloadstory=None, prompt='', quantkv=0, quiet=False, ratelimit=0, remotetunnel=False, ropeconfig=[0.0, 10000.0], savedatafile=None, sdclamped=0, sdclampedsoft=0, sdclip1='U:/z-image turbo/Qwen_3_4b-FP16.gguf', sdclip2='', sdclipgpu=False, sdconfig=None, sdconvdirect='off', sdflashattention=False, sdgendefaults='', sdlora='', sdloramult=1.0, sdmodel='C:/Users/Nitra/Downloads/z_image_turbo-Q4_0.gguf', sdnotile=False, sdoffloadcpu=False, sdphotomaker='', sdquant=0, sdt5xxl='', sdthreads=15, sdtiledvae=768, sdvae='C:/Users/Nitra/Downloads/flux2vae.safetensors', sdvaeauto=False, sdvaecpu=False, showgui=False, singleinstance=False, skiplauncher=False, smartcontext=False, ssl=None, tensor_split=None, testmemory=False, threads=15, ttsgpu=False, ttsmaxlen=4096, ttsmodel='', ttsthreads=0, ttswavtokenizer='', unpack='', useclblast=None, usecpu=False, usecuda=['normal', '0', 'mmq'], usemlock=False, usemmap=False, useswa=False, usevulkan=None, version=False, visionmaxres=1024, websearch=False, whispermodel='')

ImageGen Init - Load Model: C:\Users\Nitra\Downloads\z_image_turbo-Q4_0.gguf
With Custom VAE: C:\Users\Nitra\Downloads\flux2vae.safetensors
With Custom Clip-1 Model: U:\z-image turbo\Qwen_3_4b-FP16.gguf

Swap to Diffusion Model Path:C:\Users\Nitra\Downloads\z_image_turbo-Q4_0.gguf
Try read vocab from C:\Users\Nitra\AppData\Local\Temp_MEI117922/embd_res/qwen2_merges_utf8_c_str.embd
|====================> | 453/1099 - 311.34it/s
|======================================> | 851/1099 - 173.21it/s
|======================================> | 851/1099 - 173.07it/s
Error: KCPP SD Failed to create context!
If using Flux/SD3.5, make sure you have ALL files required (e.g. VAE, T5, Clip...) or baked in!
Otherwise, if you are using GGUF format, you can try the original .safetensors instead (Comfy GGUF not supported)
Load Image Model OK: False

Error: Could not load image model: C:\Users\Nitra\Downloads\z_image_turbo-Q4_0.gguf

[wbruna]

flux2vae.safetensors

You need the Flux1 VAE; the one from Flux2 doesn't work.

@LostRuins , maybe we should add a specific warning about it?

[chaos4u]

I got a working config, at the moment . looking at the presetfile in the link it specifies the following files.

z_image_turbo-Q4_0.gguf
Qwen3-4B-Instruct-2507-Q4_K_S.gguf
ae.safetensors

unlike the indidual links. which want the flux1 vae, and a mistral model.

i think it was the vae file that did the trick.

ill test a couple other configs to see if its exclusively the vae throwing off the loading of the model.

[LostRuins]

unlike the indidual links. which want the flux1 vae, and a mistral model.

that's for loading of Flux2, which is a completely unrelated model that just happens to be in the same release. i'll try to make it clearer