Qwen3.8-27B-TURBO-Fable · oQ8-fp16-mtp — a Novaeon.Studio build

A vision-capable, hard-reasoning 27B dense model for Apple Silicon — with an MTP head that works under 8-bit. This is an oMLX oQ8 (near-uniform 8-bit, sensitive tensors fp16) quantisation of DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU (Apache-2.0), a multi-stage Cold-Fusion + Fable-Fusion merge, repacked for the Mac in oMLX-native format. Built, tuned, and benchmarked on an Apple M5 Max (128 GB) by Novaeon.Studio.

Why this build exists: we wanted a local model that is (a) a strong reasoner, (b) multimodal, and (c) fast enough to be usable as a dense 27B on a Mac. The native MTP head is the lever for (c) — and unlike most 8-bit repacks, this one's MTP survived quantisation and more than doubles decode. Everything in the "measured" section below is from our hardware; everything else is attributed.

---

Best for & why it's here

Intended use: research and private evaluation — creative writing, roleplay, security research and vision tasks. Apply your own safety layer before production use. A companion model, not a throughput workhorse.

Why we published it: a 27B with an intact vision tower and a working MTP head (+~115% decode). It's dense, so slower than our A3B seat (~26 tok/s) — we ship it as the selectable vision model rather than the default agent.

Highlights

• Behaviour inherited from the base. The base card reports a low decline rate on its own safety-prompt evaluation (0/100 declined) at KL divergence 0.0535; our own probes on these quantised weights also returned zero declined requests.
• Working MTP → +115% decode. The native multi-token-prediction head is not an inert graft: under oQ8 it sustains real draft acceptance. 26.2 tok/s with draft-6 vs 12.2 tok/s with MTP off.
• Top-of-class reasoning for its size. The base is the first model of its size/type to breach ARC-C 0.735 in 8-bit (and 0.719 in 4-bit) — well clear of the Qwen3.5/3.6/3.8 27B-class baselines.
• "TURBO" thinking budget. The base's TURBO treatment cuts reasoning-token spend to roughly 1/2 – 1/10 of a standard thinking model, which is what makes a 26 tok/s dense 27B practical for interactive use.
• Vision intact. Multimodal (text + image → text); the vision tower is preserved from the base.

---

Measured on our hardware (Apple M5 Max 128 GB · oMLX)

MTP speculative decoding

_{Standalone run (our serving seat unloaded), 220-token generation, thinking off. +115% over MTP-off — i.e. the head is doing real speculative work under 8-bit, not idling.}

Scope note (honest): these throughput numbers and the safety-prompt probes are ours. We did not run full IFEval / agentic suites against this build — so we don't quote any. The accuracy numbers below belong to the base model and are attributed as such.

---

Base-model benchmarks

In-class ARC-Challenge comparison (mxfp8), benchmarks by Nightmedia, measured on the base model:

Full seven-benchmark sweep on the base:

_{Reproduced from the base model card / Nightmedia benchmark runs. These were measured on the base weights, not on this oQ8 repack.}

Other base-card highlights (credit DavidAU): multi-stage Cold-Fusion + Fable-Fusion-711 fine-tune/merge; three reasoning modes (xhigh — the default — plus medium and low); "TURBO" reduced-thinking-token treatment.

---

Optimal oMLX settings (figured out empirically)

We swept the tuning levers on this build. Because the native MTP head genuinely accelerates decode under oQ8, the optimal profile is MTP-on with ANE-prefill off — with a working MTP, ANE-prefill's dispatch overhead is a net loss.

Drop-in ~/.omlx/model_settings.json entry

{
  "max_context_window": 262144,
  "mtp_enabled": true,
  "mtp_num_draft_tokens": 6,
  "qwen35_ane_prefill_enabled": false,
  "turboquant_kv_enabled": true,
  "turboquant_kv_bits": 8.0,
  "turboquant_skip_last": true,
  "qwen35_oq_a8_enabled": false,
  "dflash_enabled": false
}

How it was built: oMLX admin oQ conversion with oq_level: 8, dtype: float16, and — critically — preserve_mtp: true. Without that flag the mtp.* tensors are dropped and the speculative head is gone; with it, all 29 MTP tensors ride along and oMLX reports mtp_compatible: true.

On the architecture: qwen3_5 dense hybrid — most layers use GatedDeltaNet linear attention with periodic full-attention layers interleaved. Only the full-attention layers carry a growing KV cache, so memory stays modest even far into the 256k window.

---

Quickstart

# oMLX (recommended, Apple Silicon)
omlx serve NovaeonStudio/Qwen3.8-27B-TURBO-Fable-Heretic-oQ8-fp16-mtp --port 8000
# then hit the OpenAI-compatible endpoint at http://127.0.0.1:8000/v1

# mlx-vlm
pip install -U mlx-vlm
python -m mlx_vlm.generate --model NovaeonStudio/Qwen3.8-27B-TURBO-Fable-Heretic-oQ8-fp16-mtp \
  --max-tokens 4096 --temperature 0.6 --top-p 0.95 \
  --prompt "A snail climbs 3m by day, slips 2m by night, in a 10m well. How many days to escape?"

Recommended inference parameters

temperature 0.6 · top_p 0.95 · top_k 20 · reasoning model (emits <think>; set enable_thinking: false for plain answers) · parse and strip <think>…</think> for end users. For long-form / creative work the base author suggests the xhigh reasoning mode (the default); drop to medium or low when you want speed over depth.

---

Intended use & limitations

The base model has had its default safety behaviour reduced by a post-training weight edit (done by the base authors, not by us). Practically:

• It is less likely to decline requests on topic alone and does not add warnings on its own.
• The base author notes it may still need an explicit instruction about tone, explicitness or format for some content.
• It is intended for research and private evaluation; apply your own safety layer before production use, and comply with the laws and platform terms that apply to you.

---

Attribution & license

• Base model: DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU — Apache-2.0, by DavidAU. Multi-stage Cold-Fusion + Fable-Fusion merge on the Qwen3.8 27B family.
• Benchmarks (ARC/etc.): Nightmedia, measured on the base model.
• This quantization & serving profile: Novaeon.Studio, 2026. Released under Apache-2.0, same as the base.

This is an independent community quantization. It is not endorsed by DavidAU, p-e-w, or the Qwen authors. All credit for the model's capabilities belongs to them.

@misc{novaeon2026qwen3827bturbofableoq8,
  title  = {Qwen3.8-27B-TURBO-Fable oQ8-fp16-mtp: an oMLX build for Apple Silicon},
  author = {Novaeon.Studio},
  year   = {2026},
  note   = {Quantization of DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU},
  url    = {https://huggingface.co/NovaeonStudio/Qwen3.8-27B-TURBO-Fable-Heretic-oQ8-fp16-mtp}
}

_{novæon — digital business architecture + AI · novaeon.studio}