github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-21 17:37:25 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #16019] gemma4:31b-coding-mtp-bf16 getting error "requires macOS" on Linux NVIDIA GDX Spark #87887

New Issue
Closed
opened 2026-05-10 06:32:06 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-05-10 06:32:06 -05:00
Owner
Copy Link
Copy Source

Originally created by @sjgiewont on GitHub (May 6, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/16019

Description:

I am attempting to pull the new gemma4:31b-coding-mtp-bf16 model on a Linux system with NVIDIA hardware. Since this is a standard architecture, it should pull and run via CUDA.

Actual Behavior:

The pull fails immediately with:

$ ollama pull gemma4:31b-coding-mtp-bf16
pulling manifest 
Error: pull model manifest: 412: this model requires macOS

$ ollama --version
ollama version is 0.23.1

System Info

May 06 10:04:33 systemd[1]: Started ollama.service - Ollama Service.
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.107-06:00 level=INFO source=routes.go:1782 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLL>
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.107-06:00 level=INFO source=routes.go:1784 msg="Ollama cloud disabled: false"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.120-06:00 level=INFO source=images.go:517 msg="total blobs: 53"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.121-06:00 level=INFO source=images.go:524 msg="total unused blobs removed: 0"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.122-06:00 level=INFO source=routes.go:1847 msg="Listening on [::]:11434 (version 0.23.1)"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.122-06:00 level=INFO source=runner.go:67 msg="discovering available GPUs..."
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.122-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 43199"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.276-06:00 level=INFO source=model_recommendations.go:179 msg="model recommendations cache sleep scheduled" wait=3h54m54.143217423s consecutive_failures=0
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.498-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 42669"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.878-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 35017"
May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.878-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 45293"
May 06 10:04:34 ollama[505222]: time=2026-05-06T10:04:34.489-06:00 level=INFO source=types.go:42 msg="inference compute" id=GPU-0ab97a9d-1079-edae-f585-30fa6e19c6a2 filter_id="" library=CUDA compute=12.1 name=CUDA0 description="NVIDIA GB10" libdirs=ollama,cuda_v13 driver=13.0 pci_>
May 06 10:04:34 ollama[505222]: time=2026-05-06T10:04:34.489-06:00 level=INFO source=routes.go:1897 msg="vram-based default context" total_vram="119.7 GiB" default_num_ctx=262144
Originally created by @sjgiewont on GitHub (May 6, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/16019 ## Description: I am attempting to pull the new `gemma4:31b-coding-mtp-bf16` model on a Linux system with NVIDIA hardware. Since this is a standard architecture, it should pull and run via CUDA. ## Actual Behavior: The pull fails immediately with: ``` $ ollama pull gemma4:31b-coding-mtp-bf16 pulling manifest Error: pull model manifest: 412: this model requires macOS $ ollama --version ollama version is 0.23.1 ``` ## System Info ``` May 06 10:04:33 systemd[1]: Started ollama.service - Ollama Service. May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.107-06:00 level=INFO source=routes.go:1782 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLL> May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.107-06:00 level=INFO source=routes.go:1784 msg="Ollama cloud disabled: false" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.120-06:00 level=INFO source=images.go:517 msg="total blobs: 53" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.121-06:00 level=INFO source=images.go:524 msg="total unused blobs removed: 0" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.122-06:00 level=INFO source=routes.go:1847 msg="Listening on [::]:11434 (version 0.23.1)" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.122-06:00 level=INFO source=runner.go:67 msg="discovering available GPUs..." May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.122-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 43199" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.276-06:00 level=INFO source=model_recommendations.go:179 msg="model recommendations cache sleep scheduled" wait=3h54m54.143217423s consecutive_failures=0 May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.498-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 42669" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.878-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 35017" May 06 10:04:33 ollama[505222]: time=2026-05-06T10:04:33.878-06:00 level=INFO source=server.go:433 msg="starting runner" cmd="/usr/local/bin/ollama runner --ollama-engine --port 45293" May 06 10:04:34 ollama[505222]: time=2026-05-06T10:04:34.489-06:00 level=INFO source=types.go:42 msg="inference compute" id=GPU-0ab97a9d-1079-edae-f585-30fa6e19c6a2 filter_id="" library=CUDA compute=12.1 name=CUDA0 description="NVIDIA GB10" libdirs=ollama,cuda_v13 driver=13.0 pci_> May 06 10:04:34 ollama[505222]: time=2026-05-06T10:04:34.489-06:00 level=INFO source=routes.go:1897 msg="vram-based default context" total_vram="119.7 GiB" default_num_ctx=262144 ```
GiteaMirror commented 2026-05-10 06:32:07 -05:00
Author
Owner
Copy Link
Copy Source

@rick-github commented on GitHub (May 6, 2026):

MTP support requires the MLX runner. The MLX runner is currently only supported on macOS.

<!-- gh-comment-id:4390127302 --> @rick-github commented on GitHub (May 6, 2026): [MTP support](https://github.com/ollama/ollama/pull/15980) requires the MLX runner. The MLX runner is currently only supported on macOS.
GiteaMirror commented 2026-05-10 06:32:07 -05:00
Author
Owner
Copy Link
Copy Source

@blknight88 commented on GitHub (May 7, 2026):

Are there plans to add the MLX runner to other operating systems besides macOS?

<!-- gh-comment-id:4399976130 --> @blknight88 commented on GitHub (May 7, 2026): Are there plans to add the MLX runner to other operating systems besides macOS?
GiteaMirror commented 2026-05-10 06:32:07 -05:00
Author
Owner
Copy Link
Copy Source

@rick-github commented on GitHub (May 7, 2026):

Are there plans to add the MLX runner to other operating systems besides macOS?

Yes. For example, Linux:

$ for i in qwen3.6:35b-a3b-bf16 qwen3.6:35b-a3b-mlx-bf16 ; do echo $i ; ollama run $i 'why is the sky blue?' --verbose 2>&1 | grep ^eval.rate ; ollama ps ; done
qwen3.6:35b-a3b-bf16
eval rate:            105.47 tokens/s
qwen3.6:35b-a3b-mlx-bf16
eval rate:            126.98 tokens/s

NAME                        ID              SIZE     PROCESSOR    CONTEXT    UNTIL   
qwen3.6:35b-a3b-bf16        94061ddd23a7    82 GB    100% GPU     262144     Forever    
qwen3.6:35b-a3b-mlx-bf16    6d7a4cae9475    69 GB    100% GPU     262144     Forever

MLX is not mature on non-macOS platforms yet so is not officially supported.

<!-- gh-comment-id:4400036482 --> @rick-github commented on GitHub (May 7, 2026): > Are there plans to add the MLX runner to other operating systems besides macOS? Yes. For example, Linux: ```console $ for i in qwen3.6:35b-a3b-bf16 qwen3.6:35b-a3b-mlx-bf16 ; do echo $i ; ollama run $i 'why is the sky blue?' --verbose 2>&1 | grep ^eval.rate ; ollama ps ; done qwen3.6:35b-a3b-bf16 eval rate: 105.47 tokens/s qwen3.6:35b-a3b-mlx-bf16 eval rate: 126.98 tokens/s ``` ``` NAME ID SIZE PROCESSOR CONTEXT UNTIL qwen3.6:35b-a3b-bf16 94061ddd23a7 82 GB 100% GPU 262144 Forever qwen3.6:35b-a3b-mlx-bf16 6d7a4cae9475 69 GB 100% GPU 262144 Forever ``` MLX is not mature on non-macOS platforms yet so is not officially supported.
GiteaMirror commented 2026-05-10 06:32:08 -05:00
Author
Owner
Copy Link
Copy Source

@pdevine commented on GitHub (May 8, 2026):

Are there plans to add the MLX runner to other operating systems besides macOS?

Yes! We're making progress, but it's still not quite ready yet though. I'm going to go ahead and close the issue though, but feel free to keep commenting.

<!-- gh-comment-id:4410371503 --> @pdevine commented on GitHub (May 8, 2026): > Are there plans to add the MLX runner to other operating systems besides macOS? Yes! We're making progress, but it's still not quite ready yet though. I'm going to go ahead and close the issue though, but feel free to keep commenting.
GiteaMirror commented 2026-05-10 06:32:09 -05:00
Author
Owner
Copy Link
Copy Source

@ruruex commented on GitHub (May 9, 2026):

Hope we can get the Nvidia GPU version

<!-- gh-comment-id:4412285746 --> @ruruex commented on GitHub (May 9, 2026): Hope we can get the Nvidia GPU version
GiteaMirror commented 2026-05-10 06:32:09 -05:00
Author
Owner
Copy Link
Copy Source

@tdbe commented on GitHub (May 9, 2026):

Ok so you're basically saying if you're on Windows, use llama.cpp and stop waiting for ollama... Finally a clear answer, I kept waiting but now that I understand that MTP will also only run on your mac engine (that you don't mention on the website / models), and that the best hope is that eventually it may also work on linux but not windows, then that's the straw that broke the camel's back. Time to migrate my server and workflows to llama.cpp. Thanks for making me waste infra effort into the ollama ecosystem. I was defending ollama for like 6 months to all the llama.cpp people that were running circles around me......

FIY most people running local llms only have their gaming rigs or laptops, as servers which aren't built to work on linux (and nobody has an llm-exclusinve mac - which is much slower than the GPU PCs anyway). "Let's focus on mac engines!!" ok buddies you do that.

<!-- gh-comment-id:4413748758 --> @tdbe commented on GitHub (May 9, 2026): Ok so you're basically saying if you're on Windows, use llama.cpp and stop waiting for ollama... Finally a clear answer, I kept waiting but now that I understand that MTP will also only run on your mac engine (that you don't mention on the website / models), and that the best hope is that eventually it may also work on linux but not windows, then that's the straw that broke the camel's back. Time to migrate my server and workflows to `llama.cpp`. Thanks for making me waste infra effort into the ollama ecosystem. I was defending ollama for like 6 months to all the llama.cpp people that were running circles around me...... FIY most people running local llms only have their gaming rigs or laptops, as servers which aren't built to work on linux (and nobody has an llm-exclusinve mac - which is much slower than the GPU PCs anyway). "Let's focus on mac engines!!" ok buddies you do that.
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 labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
GiteaMirror ninjasurge (Kaehlen Rog)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#87887