Core AI is Apple's on-device ML runtime in iOS 27 / macOS 27 and the successor to Core ML: PyTorch models are exported with Apple's
coreai-torch (LLMs:
coreai.llm.export) into
.aimodel bundles that run on the GPU or the Neural Engine, e.g. Qwen3-8B 4-bit decodes at 94 tok/s on an M4 Max GPU, MLX 90 under the same protocol (
apple-silicon-llm-bench, macOS 27 beta, 2026-06).
GLM-Image — Core AI
ZhipuAI's GLM-Image (16B, MIT) converted to
Core AI for on-device image generation on Apple Silicon (macOS 27+) — the zoo's first
autoregressive + diffusion hybrid.
GLM-Image generates in two stages: a 9B GLM-4 autoregressive model writes the image as a
grid of discrete visual prior tokens (sampled left-to-right like an LLM, ~36 tok/s on
M4 Max), and a 7B flow-matching DiT denoises the actual pixels conditioned on those
tokens, followed by a 16-channel VAE decode. Both native 1024×1024 and a faster
512×512 variant are included.
macOS only. The AR (9.6 GB) + DiT (9.2 GB) weights are resolution-independent, so the
512 variant is no smaller — an iPhone port needs int4 + sequential component loading.
Components
| Component | Description | Size |
|---|
glm_image_ar.aimodelc | GLM-4-9B visual-token AR decoder (int8, S=1 decode, AOT h16c GPU) | 9.6 GB |
glm_image_dit_1024.aimodelc | 30-block flow-matching DiT, 1024² (int8, AOT h16c GPU) | 9.2 GB |
glm_image_dit_512.aimodelc | Same DiT exported at 512² | 9.2 GB |
glm_image_vae_1024.aimodel | 16ch AutoencoderKL decoder, 1024² (fp32, CPU) | 0.9 GB |
glm_image_vae_512.aimodel | Same VAE exported at 512² | 0.9 GB |
tokenizer/ | GLM tokenizer (vocab 168 064; visual tokens < 16 512) | — |
ehs.f32 | Empty-glyph text embedding [1×1×1472] — the DiT text input for glyph-free prompts (no T5 needed at runtime) | 6 KB |
Weights are int8 (per-block-32) for AR/DiT; the VAE stays fp32 (fp16 overflows its
activations). Compute precision float16 on GPU. One resolution needs ~19.7 GB on disk
(AR + one DiT + one VAE).
Usage
Sample app (easiest)
CoreAIImageGen (macOS)
— run the
CoreAIImageGenMac scheme, pick
GLM-Image 1024 (AR+diffusion) from the model
menu, tap
Download & Load, type a prompt,
Generate.
- Steps 20 (default) ≈ 4.5 min/image on M4 Max — reference quality
- Steps 12 ≈ 2.5 min — nearly indistinguishable
- The 512 variant generates in ~1 min
The AR stage is sampled (temperature 0.9 / top-p 0.75, the upstream generation_config),
so re-rolling the seed meaningfully changes composition.
Pipeline contract (build your own)
The hybrid does not fit Apple's high-level CoreAIDiffusionPipeline; the host loop is
~300 lines of Swift (see the app's GlmImagePipeline):
- AR:
tokenize(prompt) + fixed grid suffix → S=1 prefill/decode with host-computed
3D-mRoPE cos/sin → sample visual tokens → 2× upsample the large grid → prior[1, N]
(N = 1024 @512, 4096 @1024).
- DiT: 20-step CFG (guidance 1.5,
prior_scale 1/0 for cond/uncond). Condition the
DiT on the raw integer schedule trunc(linspace(1000,1,steps+1)) − 1; step the
latent with the mu-shifted sigmas σ' = μ/(μ + (1/σ − 1)), μ = 0.75·side/256 + 0.25.
- VAE (fp32, CPU): latent → RGB.
⚠️ Feeding the DiT the shifted timesteps (sigma×1000) instead of the raw schedule skews
the adaLN time conditioning every step and drifts colors — resolution-dependently (mild
at 512, strong at 1024). This is the one non-obvious contract in the port.
Limitations
- Glyph rendering (text in images) is not wired: prompts containing quoted text
(
'…', "…", 「…」) route through a T5 glyph encoder that is not part of this bundle.
Keep prompts glyph-free.
- Negative prompts are a no-op (GLM's CFG drops the prior tokens instead).
- Image-to-image / editing (
GlmImageKVCache) not yet exported.
Quality
Verified against the bf16 diffusers reference: DiT single-forward cosine 0.9999, VAE
byte-exact, AR prior-injection reproduces the reference composition; end-to-end output is
visually on par with GlmImagePipeline at the same step count.
License
MIT (inherited from
zai-org/GLM-Image).
More models in this format: Core AI Model Zoo — 75 models, each with the recipe that produced it.
Want a different model on-device? Open a request — free, open weights only; the export and its measured numbers get published publicly.
