Qwen3.6-35B-A3B — Core AI (
gather_qmm
kernel, 2.1× faster)
Apple
Core AI
(
.aimodel
) conversion of
Qwen/Qwen3.6-35B-A3B
(text decoder): Qwen3.5's hybrid GatedDeltaNet + gated-attention body with a
256-expert top-8
sparse MoE
(+ shared expert). 35B total /
~3B active per token
.
Apple's
GatherMM
composite gathers the routed experts then runs a
dense matmul that reads all
256 experts' weights every token
— over-read-bound at 30.9 tok/s. This bundle uses a custom
coreai_torch.TorchMetalKernel
that takes the routed indices as a kernel input and reads
only the
8 routed experts'
weight slabs (8/256), so decode runs at active-param (~3B) bandwidth:
64.9
tok/s, 2.1×
.
Quality is clean and unchanged.
The kernel reads the
sym8
scheme = the same
symmetric-linear int8 (per-K-block-32) recipe the standard int8 bundle uses, via a
bit-exact
gather:
0 introduced flips / 18 vs fp16
(the shipped GatherMM int8 was 14/16 vs the bf16 oracle;
this matches it). So this is a pure speed win at the same quality.
bundle
size
decode tok/s
quality
gpu-pipelined/qwen3_6_35b_a3b_decode_sym8_gather/
35 GB
64.9
clean (0 flips/18 vs fp16) ✅
Mac-only (35 GB int8 is far past the iPhone limit; this is the 64/128 GB-Mac flagship).
Qwen3.6-35B-A3B-CoreAI huggingface.co is an AI model on huggingface.co that provides Qwen3.6-35B-A3B-CoreAI's model effect (), which can be used instantly with this mlboydaisuke Qwen3.6-35B-A3B-CoreAI model. huggingface.co supports a free trial of the Qwen3.6-35B-A3B-CoreAI model, and also provides paid use of the Qwen3.6-35B-A3B-CoreAI. Support call Qwen3.6-35B-A3B-CoreAI model through api, including Node.js, Python, http.
Qwen3.6-35B-A3B-CoreAI huggingface.co is an online trial and call api platform, which integrates Qwen3.6-35B-A3B-CoreAI's modeling effects, including api services, and provides a free online trial of Qwen3.6-35B-A3B-CoreAI, you can try Qwen3.6-35B-A3B-CoreAI online for free by clicking the link below.
mlboydaisuke Qwen3.6-35B-A3B-CoreAI online free url in huggingface.co:
Qwen3.6-35B-A3B-CoreAI is an open source model from GitHub that offers a free installation service, and any user can find Qwen3.6-35B-A3B-CoreAI on GitHub to install. At the same time, huggingface.co provides the effect of Qwen3.6-35B-A3B-CoreAI install, users can directly use Qwen3.6-35B-A3B-CoreAI installed effect in huggingface.co for debugging and trial. It also supports api for free installation.
Qwen3.6-35B-A3B-CoreAI install url in huggingface.co: