GLM-5.3-Flash on a Mac: Which Studio Runs the Ox Alpha Model (2026)

What GLM-5.3-Flash is

On 26 August 2026 Zhipu confirmed that the model circulating as “Ox Alpha” was GLM-5.3-Flash and published the weights on Hugging Face under MIT — the most permissive license in the index, scored 10/10.

From the model card: 320B total parameters, 18B active per token, a hybrid of sparse and linear attention with what Zhipu calls manifold-constrained hyper-connections, native image input alongside text, and a 300K-token evaluation context. It is the first natively multimodal model in the GLM-5 series. The card reports Terminal-Bench 2.1 at 84.3; Zhipu’s launch line is that it outperforms GLM-5.2 across benchmarks at a tenth of the price and approaches Claude Opus 4.8 on coding and agentic tasks. No SWE-Bench Pro number appears on the card as of 7 September, so the index scores capability at 48 on the evidence available — one point under GLM-5.2, which it has not yet been shown to beat on a standardized harness.

Its sibling shipped alongside it: the 753B GLM-5.3 flagship, also open-weight since 25 August but under a custom license rather than MIT, and a 256 GB Studio proposition at 2-bit. This post is about the one you can run on 128 GB.

Which Mac holds which quant

Unsloth’s dynamic GGUFs are the practical reference. The sizes below are theirs; the Mac tiers are ours.

The line that matters: 128 GB is the floor, and 256 GB is where you stop compromising. A 96 GB Mac cannot hold any of these.

Estimated speed per chip

These are estimates from the index’s published bandwidth model with its MoE factor applied (0.42 of the active-bytes formula, fitted to the one community-measured MoE run in the index). They are not measurements, and they will be replaced by measurements as they are linked.

For comparison, DeepSeek V4 Flash — 13B active, the closest sibling in the index — is community-measured at 39 tok/s on an M5 Max at 2-bit. GLM-5.3-Flash at 3-bit with 18B active landing in the high twenties is consistent with that; if an early run comes in well above, the index will happily be wrong in that direction. Rows live on the benchmarks page with their labels.

How to run it today

Two paths, neither of them a one-line ollama run yet as of 7 September.

llama.cpp with the GGUFs

The Unsloth GGUFs require a GLM-5-aware build. Their guide points at a specific branch rather than a tagged release; on a Mac, build with Metal on (the default) and CUDA off. Expect this to land upstream, at which point the stock brew install llama.cpp will do.

git clone --branch glm5next/upstream https://github.com/ggml-org/llama.cpp
cd llama.cpp && cmake -B build -DGGML_CUDA=OFF && cmake --build build --config Release -j
./build/bin/llama-cli -hf unsloth/GLM-5.3-Flash-GGUF:UD-IQ3_XXS -c 16384

MLX

Community MLX conversions at 2–8 bit are on Hugging Face. They are the more natural fit for an Ultra and avoid the branch dance, at the cost of being community-maintained; check that the conversion’s bit-width matches the Mac tier above before downloading 200 GB.

Should you buy a Mac for it?

Not on this model alone — but this model is the clearest case yet that the 256 GB Mac Studio tier has a job. If you already own a 128 GB machine, the 3-bit quant is worth an evening. If you are deciding between 128 and 256 GB, GLM-5.3-Flash at true 4-bit, Qwen3.8-Flash-Next with room to spare, and DeepSeek V4 Flash at 4-bit are the three models that only the larger tier runs properly, and all three arrived in the same fortnight. The M5 Ultra Studio ships on 22 September; the index’s per-Mac rankings are on the 256 GB page.

Frequently Asked Questions