Qwen3.6-35B-A3B-MXFP8-MTP

Qwen3.6-35B-A3B quantized to native MXFP8 for Apple Silicon, with the vision tower and the native Multi-Token-Prediction head preserved and enabled.

Quantization

8-bit affine linears via MLX-native mx.quantize (mode="mxfp8", group_size=32). Norms, router gates, expert biases and the full vision tower are kept in fp16 passthrough (643 passthrough tensors). MTP linears are quantized to MXFP8; MTP norm/control tensors stay fp16.

Multi-Token Prediction

This bundle keeps Qwen3.6's native MTP module and runs it as a self-speculative draft head: the MTP head proposes tokens that the main model verifies in a single pass, so decoded output stays bit-identical to plain autoregressive decoding — only faster.

Recorded on an M5 Max (vMLX runtime, 96-token deterministic prompt, output verified equal to baseline at every depth):

With vMLX prefix/KV cache layers enabled the speedup holds — a recorded cache-on A/B measured 59.8 → 88.8 tok/s (1.48×).

Absolute tok/s depends on free memory and system load. The speedup ratio — baseline vs. MTP measured back-to-back under identical conditions — is the stable figure.

Vision, MTP and caching together

These bundles run image/video input, native MTP speculative decode and prefix/KV caching in the same session — a combination not every MTP-enabled Qwen build exposes. A recorded VL probe (2026-05-16) confirms a color identification image prompt returns the correct answer through the combined MTP + VL runtime.

Loading

Loads via stock MLX tooling on Apple Silicon — the mxfp8 weights are native mx.quantize affine, no JANG runtime required for the core model.

from mlx_vlm import load, generate
model, processor = load("OsaurusAI/Qwen3.6-35B-A3B-MXFP8-MTP")

The MTP draft path is exercised by an MTP-aware runtime (vMLX); other runtimes load and decode the main model normally and ignore the MTP head.

Variants

Credits

• Quantization toolchain: JANG by Jinho Jang <eric@osaurus.ai>
• Base model: Qwen3.6-35B-A3B by Qwen