github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #15496] Ollama 0.20.5 crashes on Apple M5 (MacBook Air, 16GB) with Metal backend init failure, even on qwen2.5:0.5b #35663

New Issue
Closed
opened 2026-04-22 20:20:12 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-22 20:20:12 -05:00
Owner
Copy Link

Originally created by @Azu126 on GitHub (Apr 11, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15496

What is the issue?

Environment

  • Device: MacBook Air M5
  • Memory: 16GB
  • Storage: 512GB
  • OS: macOS
  • Ollama version: 0.20.5
  • Architecture: arm64

Problem
Ollama fails to run even very small models such as qwen2.5:0.5b.

I first noticed the issue after trying Qwen3.5 35B A3B, but now the same error happens even with qwen2.5:0.5b.
This does not look like a memory issue, because the same failure occurs with a 0.5B model.

Reproduction

ollama run qwen2.5:0.5b

Also reproduced with:

pkill ollama
ollama serve
# in another terminal
ollama run qwen2.5:0.5b

Observed behavior
The request fails with:

Error: 500 Internal Server Error: llama runner process has terminated: %!w(<nil>)

Server logs show Metal backend initialization failure.

Additional info
	•	ollama --version works normally
	•	ollama list works normally
	•	Removing and pulling qwen2.5:0.5b again did not help
	•	Restarting Ollama did not help
	•	ollama ps is empty after failure
	•	The model file appears to load, but the runner crashes when initializing Metal
	•	Logs mention:
	•	GPU name: Apple M5
	•	the device does not have a precompiled Metal library - this is unexpected
	•	will try to compile it on the fly

Expectation
At minimum, qwen2.5:0.5b should run on this machine.
Instead, Ollama crashes during Metal backend initialization.

Question
Is this a known compatibility issue with Apple M5 / Metal in Ollama 0.20.5?
Is there any workaround, such as disabling the problematic Metal path completely?


### Relevant log output

```shell
ggml_metal_library_init: error: Error Domain=MTLLibraryErrorDomain Code=3
...
static_assert failed due to requirement '__tensor_ops_detail::__is_same_v<bfloat, half>' "Input types must match cooperative tensor types"
...
static_assert failed due to requirement '__tensor_ops_detail::__is_same_v<half, bfloat>' "Input types must match cooperative tensor types"
...
ggml_metal_init: error: failed to initialize the Metal library
ggml_backend_metal_device_init: error: failed to allocate context
llama_init_from_model: failed to initialize the context: failed to initialize Metal backend
panic: unable to create llama context
...
llama runner terminated
...
Load failed ... error="llama runner process has terminated: %!w(<nil>)"

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

Version 0.20.5 (0.20.5)

Originally created by @Azu126 on GitHub (Apr 11, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15496 ### What is the issue? **Environment** - Device: MacBook Air M5 - Memory: 16GB - Storage: 512GB - OS: macOS - Ollama version: 0.20.5 - Architecture: arm64 **Problem** Ollama fails to run even very small models such as `qwen2.5:0.5b`. I first noticed the issue after trying `Qwen3.5 35B A3B`, but now the same error happens even with `qwen2.5:0.5b`. This does not look like a memory issue, because the same failure occurs with a 0.5B model. **Reproduction** ```bash ollama run qwen2.5:0.5b Also reproduced with: pkill ollama ollama serve # in another terminal ollama run qwen2.5:0.5b Observed behavior The request fails with: Error: 500 Internal Server Error: llama runner process has terminated: %!w(<nil>) Server logs show Metal backend initialization failure. Additional info • ollama --version works normally • ollama list works normally • Removing and pulling qwen2.5:0.5b again did not help • Restarting Ollama did not help • ollama ps is empty after failure • The model file appears to load, but the runner crashes when initializing Metal • Logs mention: • GPU name: Apple M5 • the device does not have a precompiled Metal library - this is unexpected • will try to compile it on the fly Expectation At minimum, qwen2.5:0.5b should run on this machine. Instead, Ollama crashes during Metal backend initialization. Question Is this a known compatibility issue with Apple M5 / Metal in Ollama 0.20.5? Is there any workaround, such as disabling the problematic Metal path completely? ### Relevant log output ```shell ggml_metal_library_init: error: Error Domain=MTLLibraryErrorDomain Code=3 ... static_assert failed due to requirement '__tensor_ops_detail::__is_same_v<bfloat, half>' "Input types must match cooperative tensor types" ... static_assert failed due to requirement '__tensor_ops_detail::__is_same_v<half, bfloat>' "Input types must match cooperative tensor types" ... ggml_metal_init: error: failed to initialize the Metal library ggml_backend_metal_device_init: error: failed to allocate context llama_init_from_model: failed to initialize the context: failed to initialize Metal backend panic: unable to create llama context ... llama runner terminated ... Load failed ... error="llama runner process has terminated: %!w(<nil>)" ``` ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version Version 0.20.5 (0.20.5)
GiteaMirror added the bug label 2026-04-22 20:20:12 -05:00
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#35663
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?