Ollama v0.32.12 adds Qwen 3.8 27B support
Ollama v0.32.12 adds support for Qwen 3.8 27B. The release provides a standard run command and an MLX variant that Ollama says is optimized for Apple Silicon devices, while leaving performance benchmarks and hardware requirements unspecified.
Ollama v0.32.12 adds support for Qwen 3.8 27B, including an MLX variant intended for Apple Silicon devices.
01
What we know now
- 01
[1] Ollama releases, v0.32.12, published 14 August 2026. https://github.com/ollama/ollama/releases/tag/v0.32.12
- 02
[1] The first-party release record states that support for Qwen 3.8 27B was added and provides commands for standard and MLX variants.
02
Why this matters for Armenia
For developers and professional users in Armenia, the update documents an additional locally runnable model option and a separate Apple Silicon-oriented variant.
03
Standard model
Listed as the general Qwen 3.8 27B command.Apple Silicon variant
Described by Ollama as an MLX variant optimized for Apple Silicon.Release
First-party support announcement.The release provides commands and an optimization claim, but no benchmark results or hardware requirements.
04
What the release adds
Ollama’s v0.32.12 release states that it adds support for Qwen 3.8 27B. This expands the set of models documented as runnable through Ollama’s local-model workflow.
The announcement is a support update rather than an independently tested performance report. The supplied record does not include comparative measurements or system specifications.
- The release identifies Qwen 3.8 27B as a newly supported model.
- Ollama provides the command `ollama run qwen3.8:27b` for the standard model option.
05
Apple Silicon option
Alongside the standard command, Ollama lists `qwen3.8:27b-mlx`, an MLX variant for Apple Silicon. The release says this variant has been optimized specifically for those devices.
Users should treat the performance and quality description as a first-party claim. The release does not state which Apple Silicon hardware configurations are required or provide benchmark evidence.
- The MLX command is `ollama run qwen3.8:27b-mlx`.
- Ollama describes this option as optimized for Apple Silicon devices.
- The release characterizes the optimization in terms of performance and output quality for repeated tasks and coding agents.
06
What users can take from the update
The practical change is a newly documented 27B model option in Ollama, with a distinct Apple Silicon-oriented path. It may be relevant to local workflows involving coding, research, and other professional work, which the release associates with the model.
The announcement alone does not establish how the model will perform for a particular workflow. Local testing remains necessary to assess speed, output quality, and device compatibility.
- Check local hardware capacity before downloading or running a 27B model.
- Use the standard command or the Apple Silicon MLX command according to the device in use.
- Evaluate suitability with the intended tasks, since no benchmark data is supplied in the release record.
07
Getting started
Ollama lists two commands for accessing the newly supported model variant.
- 01
Run `ollama run qwen3.8:27b` to use Qwen 3.8 27B.
- 02
On Apple Silicon devices, run `ollama run qwen3.8:27b-mlx` for the MLX variant.
08
Limits of this edition
This information is based on Ollama’s first-party release record.
The supplied release record does not independently verify model performance.
No benchmarks, memory requirements, supported macOS versions, or other hardware requirements are provided.
SRC
Source desk
Direct links to the material behind this selection. Seeing the source matters as much as reading the synthesis.