github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #13746] new: Image generation models (experimental) - not working #9010

New Issue
Open
opened 2026-04-12 21:50:06 -05:00 by GiteaMirror · 15 comments
GiteaMirror commented 2026-04-12 21:50:06 -05:00
Owner
Copy Link

Originally created by @leder11011 on GitHub (Jan 16, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/13746

Originally assigned to: @jmorganca on GitHub.

What is the issue?

Hello all,
thank you for ollama v0.14.1!
unfortunately I get an error on linux w/ nvidia GPU or CPU:

OLLAMA_NUM_CTX=2048 ollama run x/z-image-turbo                                                             
>>> cat in the rain

Error: 500 Internal Server Error: image runner failed: 2026/01/16 09:00:39 runner.go:65: INFO starting image runner model=x/z-image-turbo port=34187 (exit: exit status 255)
>>> /bye


  Model
    architecture    ZImagePipeline
    parameters      10.3B
    quantization    FP8
    requires        0.14.0

  Capabilities
    image

Relevant log output

Jan 16 09:02:35 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:35.999+01:00 level=INFO source=server.go:149 msg="starting ollama-mlx image runner subprocess" exe=/usr/local/bin/ollama-mlx model=x/z-image-turbo port=36725
Jan 16 09:02:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:36.171+01:00 level=WARN source=server.go:141 msg=image-runner msg="2026/01/16 09:02:36 runner.go:65: INFO starting image runner model=x/z-image-turbo port=36725"
Jan 16 09:02:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:36.174+01:00 level=INFO source=server.go:134 msg=image-runner msg="Loading Z-Image model from manifest: x/z-image-turbo..."
Jan 16 09:02:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:36.439+01:00 level=INFO source=server.go:134 msg=image-runner msg="  Loading tokenizer... ✓"
Jan 16 09:02:37 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:37.756+01:00 level=INFO source=server.go:134 msg=image-runner msg="  Loading text encoder... ✓"
Jan 16 09:02:38 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:38.722+01:00 level=INFO source=server.go:134 msg=image-runner msg="  (11.3 GB, peak 11.3 GB)"
Jan 16 09:02:40 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:40.936+01:00 level=INFO source=server.go:134 msg=image-runner msg="  Loading transformer... ✓"
Jan 16 09:04:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:04:36.000+01:00 level=INFO source=server.go:320 msg="stopping image runner subprocess" pid=27349
Jan 16 09:04:36 gerrit-systemproductname ollama[17756]: [GIN] 2026/01/16 - 09:04:36 | 500 |          2m0s |       127.0.0.1 | POST     "/api/generate"

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @leder11011 on GitHub (Jan 16, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/13746 Originally assigned to: @jmorganca on GitHub. ### What is the issue? Hello all, thank you for ollama v0.14.1! unfortunately I get an error on linux w/ nvidia GPU or CPU: ``` OLLAMA_NUM_CTX=2048 ollama run x/z-image-turbo >>> cat in the rain Error: 500 Internal Server Error: image runner failed: 2026/01/16 09:00:39 runner.go:65: INFO starting image runner model=x/z-image-turbo port=34187 (exit: exit status 255) >>> /bye Model architecture ZImagePipeline parameters 10.3B quantization FP8 requires 0.14.0 Capabilities image ``` ### Relevant log output ```shell Jan 16 09:02:35 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:35.999+01:00 level=INFO source=server.go:149 msg="starting ollama-mlx image runner subprocess" exe=/usr/local/bin/ollama-mlx model=x/z-image-turbo port=36725 Jan 16 09:02:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:36.171+01:00 level=WARN source=server.go:141 msg=image-runner msg="2026/01/16 09:02:36 runner.go:65: INFO starting image runner model=x/z-image-turbo port=36725" Jan 16 09:02:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:36.174+01:00 level=INFO source=server.go:134 msg=image-runner msg="Loading Z-Image model from manifest: x/z-image-turbo..." Jan 16 09:02:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:36.439+01:00 level=INFO source=server.go:134 msg=image-runner msg=" Loading tokenizer... ✓" Jan 16 09:02:37 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:37.756+01:00 level=INFO source=server.go:134 msg=image-runner msg=" Loading text encoder... ✓" Jan 16 09:02:38 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:38.722+01:00 level=INFO source=server.go:134 msg=image-runner msg=" (11.3 GB, peak 11.3 GB)" Jan 16 09:02:40 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:02:40.936+01:00 level=INFO source=server.go:134 msg=image-runner msg=" Loading transformer... ✓" Jan 16 09:04:36 gerrit-systemproductname ollama[17756]: time=2026-01-16T09:04:36.000+01:00 level=INFO source=server.go:320 msg="stopping image runner subprocess" pid=27349 Jan 16 09:04:36 gerrit-systemproductname ollama[17756]: [GIN] 2026/01/16 - 09:04:36 | 500 | 2m0s | 127.0.0.1 | POST "/api/generate" ``` ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the bug label 2026-04-12 21:50:06 -05:00
GiteaMirror commented 2026-04-12 21:50:07 -05:00
Author
Owner
Copy Link

@1101728133 commented on GitHub (Jan 16, 2026):

Me too.

<!-- gh-comment-id:3758704430 --> @1101728133 commented on GitHub (Jan 16, 2026): Me too.
GiteaMirror commented 2026-04-12 21:50:08 -05:00
Author
Owner
Copy Link

@Sekousuke commented on GitHub (Jan 16, 2026):

The computer's memory is insufficient.

<!-- gh-comment-id:3758782559 --> @Sekousuke commented on GitHub (Jan 16, 2026): The computer's memory is insufficient.
GiteaMirror commented 2026-04-12 21:50:09 -05:00
Author
Owner
Copy Link

@leder11011 commented on GitHub (Jan 16, 2026):

I got a different error when omitting OLLAMA_NUM_CTX=2048

<!-- gh-comment-id:3761151829 --> @leder11011 commented on GitHub (Jan 16, 2026): I got a different error when omitting `OLLAMA_NUM_CTX=2048`
GiteaMirror commented 2026-04-12 21:50:09 -05:00
Author
Owner
Copy Link

@Sekousuke commented on GitHub (Jan 17, 2026):

OLLAMA_NUM_CTX is used for the text model, not for the image model. Moreover, it was terminated by the MacOS system, not because you restricted it.

<!-- gh-comment-id:3762664837 --> @Sekousuke commented on GitHub (Jan 17, 2026): OLLAMA_NUM_CTX is used for the text model, not for the image model. Moreover, it was terminated by the MacOS system, not because you restricted it.
GiteaMirror commented 2026-04-12 21:50:09 -05:00
Author
Owner
Copy Link

@ghmer commented on GitHub (Jan 18, 2026):

I can confirm that image generation works with both models (fp8/bp16) if the appropriate system resources are available.

<!-- gh-comment-id:3765482914 --> @ghmer commented on GitHub (Jan 18, 2026): I can confirm that image generation works with both models (fp8/bp16) if the appropriate system resources are available.
GiteaMirror commented 2026-04-12 21:50:10 -05:00
Author
Owner
Copy Link

@vansatchen commented on GitHub (Jan 18, 2026):

I can confirm that image generation works with both models (fp8/bp16) if the appropriate system resources are available.

ollama run x/z-image-turbo
a cat

Error: 500 Internal Server Error: image runner exited unexpectedly: exit status 255

ollama -v
ollama version is 0.14.1

free -h
total used free shared buff/cache available
Mem: 1.0Ti 31Gi 8.6Gi 182Mi 974Gi 976Gi
Swap: 8.0Gi 1.8Mi 8.0Gi

Is this memory enough?

<!-- gh-comment-id:3765595969 --> @vansatchen commented on GitHub (Jan 18, 2026): > I can confirm that image generation works with both models (fp8/bp16) if the appropriate system resources are available. > ollama run x/z-image-turbo > a cat > > Error: 500 Internal Server Error: image runner exited unexpectedly: exit status 255 > ollama -v > ollama version is 0.14.1 > free -h > total used free shared buff/cache available > Mem: 1.0Ti 31Gi 8.6Gi 182Mi 974Gi 976Gi > Swap: 8.0Gi 1.8Mi 8.0Gi Is this memory enough?
GiteaMirror commented 2026-04-12 21:50:10 -05:00
Author
Owner
Copy Link

@ghmer commented on GitHub (Jan 18, 2026):

I can confirm that image generation works with both models (fp8/bp16) if the appropriate system resources are available.

ollama run x/z-image-turbo
a cat
Error: 500 Internal Server Error: image runner exited unexpectedly: exit status 255

ollama -v
ollama version is 0.14.1

free -h
total used free shared buff/cache available
Mem: 1.0Ti 31Gi 8.6Gi 182Mi 974Gi 976Gi
Swap: 8.0Gi 1.8Mi 8.0Gi

Is this memory enough?

I mean, this is plenty, but is it a MacOS system?

<!-- gh-comment-id:3765603829 --> @ghmer commented on GitHub (Jan 18, 2026): > > I can confirm that image generation works with both models (fp8/bp16) if the appropriate system resources are available. > > > ollama run x/z-image-turbo > > a cat > > Error: 500 Internal Server Error: image runner exited unexpectedly: exit status 255 > > > ollama -v > > ollama version is 0.14.1 > > > free -h > > total used free shared buff/cache available > > Mem: 1.0Ti 31Gi 8.6Gi 182Mi 974Gi 976Gi > > Swap: 8.0Gi 1.8Mi 8.0Gi > > Is this memory enough? I mean, this is plenty, but is it a MacOS system?
GiteaMirror commented 2026-04-12 21:50:11 -05:00
Author
Owner
Copy Link

@vansatchen commented on GitHub (Jan 19, 2026):

I mean, this is plenty, but is it a MacOS system?

It's Linux(Ubuntu 24.04), but realese notes of v0.14.1:

Image generation models (experimental)

Experimental image generation models are available for macOS and Linux (CUDA) in Ollama:
Available models

Z-Image-Turbo

<!-- gh-comment-id:3766249571 --> @vansatchen commented on GitHub (Jan 19, 2026): > I mean, this is plenty, but is it a MacOS system? It's Linux(Ubuntu 24.04), but realese notes of [v0.14.1](https://github.com/ollama/ollama/releases/tag/v0.14.1): > Image generation models (experimental) > > Experimental image generation models are available for macOS **and Linux** (CUDA) in Ollama: > Available models > > [Z-Image-Turbo](https://ollama.com/x/z-image-turbo)
GiteaMirror commented 2026-04-12 21:50:11 -05:00
Author
Owner
Copy Link

@ghmer commented on GitHub (Jan 19, 2026):

And according to the README of the model you linked to:

Note: image generation models only work on macOS.

I guess one of these docs is not right, but given that I can run both models on my Macbook, I'd further guess - at least this model - only works on MacOS. Aren't they even relying on the MLX framework?

<!-- gh-comment-id:3768729501 --> @ghmer commented on GitHub (Jan 19, 2026): And according to the README of the model you linked to: > Note: image generation models only work on macOS. I guess one of these docs is not right, but given that I can run both models on my Macbook, I'd further guess - at least this model - only works on MacOS. Aren't they even relying on the MLX framework?
GiteaMirror commented 2026-04-12 21:50:11 -05:00
Author
Owner
Copy Link

@BOTO145 commented on GitHub (Jan 23, 2026):

the image generation will come later in windows if your os is windows (even lates)

<!-- gh-comment-id:3790573792 --> @BOTO145 commented on GitHub (Jan 23, 2026): the image generation will come later in windows if your os is windows (even lates)
GiteaMirror commented 2026-04-12 21:50:12 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 8, 2026):

Linux docker: https://github.com/ollama/ollama/issues/14016#issuecomment-3831904450
Linux native: https://github.com/ollama/ollama/issues/14046#issuecomment-3846867534
Only bf16 quants supported for flux, minimum memory 16G: https://github.com/ollama/ollama/issues/14046#issuecomment-3847105340

<!-- gh-comment-id:3868256335 --> @rick-github commented on GitHub (Feb 8, 2026): Linux docker: https://github.com/ollama/ollama/issues/14016#issuecomment-3831904450 Linux native: https://github.com/ollama/ollama/issues/14046#issuecomment-3846867534 Only bf16 quants supported for flux, minimum memory 16G: https://github.com/ollama/ollama/issues/14046#issuecomment-3847105340
GiteaMirror commented 2026-04-12 21:50:12 -05:00
Author
Owner
Copy Link

@BOTO145 commented on GitHub (Feb 9, 2026):

Image

either this is the reasons OR if you have downloaded it from hugging face, then u dont have enough vram (that's what I think)

<!-- gh-comment-id:3871531272 --> @BOTO145 commented on GitHub (Feb 9, 2026): <img width="771" height="124" alt="Image" src="https://github.com/user-attachments/assets/943eabe0-b5f6-4152-a18f-586461710261" /> either this is the reasons OR if you have downloaded it from hugging face, then u dont have enough vram (that's what I think)
GiteaMirror commented 2026-04-12 21:50:12 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 9, 2026):

Image generation works on Linux using the fixes shown in https://github.com/ollama/ollama/issues/13746#issuecomment-3868256335

<!-- gh-comment-id:3872504518 --> @rick-github commented on GitHub (Feb 9, 2026): Image generation works on Linux using the fixes shown in https://github.com/ollama/ollama/issues/13746#issuecomment-3868256335
GiteaMirror commented 2026-04-12 21:50:12 -05:00
Author
Owner
Copy Link

@ckuethe commented on GitHub (Feb 9, 2026):

For the benefit of other linux/docker/rocm users, that looks like it depends on CUDA.

It looks like MLX doesn'twith ROCm yet, based on https://github.com/ml-explore/mlx/issues/2556 and https://github.com/ml-explore/mlx/pull/2300

<!-- gh-comment-id:3873107450 --> @ckuethe commented on GitHub (Feb 9, 2026): For the benefit of other linux/docker/rocm users, that looks like it depends on CUDA. It looks like MLX doesn'twith ROCm yet, based on https://github.com/ml-explore/mlx/issues/2556 and https://github.com/ml-explore/mlx/pull/2300
GiteaMirror commented 2026-04-12 21:50:13 -05:00
Author
Owner
Copy Link

@Geramy commented on GitHub (Apr 5, 2026):

For the benefit of other linux/docker/rocm users, that looks like it depends on CUDA.

It looks like MLX doesn'twith ROCm yet, based on ml-explore/mlx#2556 and ml-explore/mlx#2300

There is a ROCm pull request waiting for some advice from the maintainers at MLX

<!-- gh-comment-id:4188005000 --> @Geramy commented on GitHub (Apr 5, 2026): > For the benefit of other linux/docker/rocm users, that looks like it depends on CUDA. > > It looks like MLX doesn'twith ROCm yet, based on [ml-explore/mlx#2556](https://github.com/ml-explore/mlx/issues/2556) and [ml-explore/mlx#2300](https://github.com/ml-explore/mlx/pull/2300) There is a ROCm pull request waiting for some advice from the maintainers at MLX
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#9010
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?