github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #14156] FP8 (F8_E4M3) tensor type support for text models #9230

New Issue
Open
opened 2026-04-12 22:05:59 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 22:05:59 -05:00
Owner
Copy Link

Originally created by @rjmalagon on GitHub (Feb 8, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14156

More than a plain feature request, it is more an inquiry about eventual FP8 support on Ollama.

Despite the limited hardware (and software) support, some models are being published on FP8 (F8_E4M3).

Notable MOE models like Qwen/Qwen3-Coder-Next-FP8 sound very welcome if FP8 retains better precision than Q8 (INT8) for the active parameters.

It is actually difficult to compare both types for an informed discussion because I only found FP8 support on VLLM. Building VLLM for my actual setup has been challenging (what a Python mess, even on containers).

Originally created by @rjmalagon on GitHub (Feb 8, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14156 More than a plain feature request, it is more an inquiry about eventual FP8 support on Ollama. Despite the limited hardware (and software) support, some models are being published on FP8 (F8_E4M3). Notable MOE models like Qwen/Qwen3-Coder-Next-FP8 sound very welcome if FP8 retains better precision than Q8 (INT8) for the active parameters. It is actually difficult to compare both types for an informed discussion because I only found FP8 support on VLLM. Building VLLM for my actual setup has been challenging (what a Python mess, even on containers).
GiteaMirror added the feature request label 2026-04-12 22:05:59 -05:00
GiteaMirror commented 2026-04-12 22:06:00 -05:00
Author
Owner
Copy Link

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

It's possible that FP8 support will come via the MLX backend, jmorgan mentioned fp8 and nvfp4 in a discord post.

<!-- gh-comment-id:3868205427 --> @rick-github commented on GitHub (Feb 8, 2026): It's possible that FP8 support will come via the MLX backend, jmorgan mentioned fp8 and nvfp4 in a [discord post](https://discord.com/channels/1128867683291627614/1464814269479780458/1464863039106056213).
GiteaMirror commented 2026-04-12 22:06:01 -05:00
Author
Owner
Copy Link

@rjmalagon commented on GitHub (Feb 8, 2026):

Sounds good @rick-github , but how far is the support of the MLX backend for Linux+ROCM or Linux+Vulkan. As far as I know, MLX is an Apple thing and only used with image generation models.

<!-- gh-comment-id:3868255046 --> @rjmalagon commented on GitHub (Feb 8, 2026): Sounds good @rick-github , but how far is the support of the MLX backend for Linux+ROCM or Linux+Vulkan. As far as I know, MLX is an Apple thing and only used with image generation models.
GiteaMirror commented 2026-04-12 22:06:02 -05:00
Author
Owner
Copy Link

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

Support for text/text models on the MLX backend is progressing, eg 0.15.5 has support for glm-4.7-flash. Currently only MacOS and Linux+CUDA are supported, but there is an open PR in ollama for Windows support, and discussion in mlx regarding the Vulkan backend (besides others).

<!-- gh-comment-id:3868341705 --> @rick-github commented on GitHub (Feb 8, 2026): Support for text/text models on the MLX backend is progressing, eg [0.15.5](https://github.com/ollama/ollama/releases/tag/v0.15.5) has support for glm-4.7-flash. Currently only MacOS and Linux+CUDA are supported, but there is an open [PR](https://github.com/ollama/ollama/pull/13806) in ollama for Windows support, and [discussion](https://github.com/ml-explore/mlx/pull/3098) in mlx regarding the Vulkan backend (besides others).
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 feature request
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#9230
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?