Downloads · 30 days
532
30% of all-time downloads
NovaeonStudio/Qwen3.8-27B-TURBO-Fable-Heretic-oQ8-fp16-mtp
Qwen3.8-27B-TURBO-Fable-Heretic-oQ8-fp16-mtp is a image-text-to-text model from NovaeonStudio. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for mlx. The card lists the license as apache-2.0.
Downloads · 30 days
532
30% of all-time downloads
All-time downloads
1.7K
Public
Parameters
27.8B
30.9 GB on disk
Likes
4
Public
Click a slice to open those files.
.safetensors30.9 GB · 100%
How the weights are stored.
U3227.3B · 98%
From the Hugging Face model README

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.
| Base model | DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU (Apache-2.0) |
| Architecture | qwen3_5 dense hybrid — GatedDeltaNet linear attention + interleaved full attention · vision encoder |
| Quantization | oQ8, near-uniform 8-bit · router/sensitive weights float16 · ~8.59 effective bpw · ~28.4 GB on disk |
| MTP head | Native, mtp_num_hidden_layers: 1 · 29 mtp.* tensors preserved in the oQ8 — and it accelerates decode |
| Alignment | Inherited from the base model, which has had its default safety behaviour reduced (see Intended use & limitations) |
| Context | 262,144 native · vision preserved |
| Engine | oMLX (Apple MLX) — VLM engine |
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.

| Configuration | Decode (tok/s) |
|---|---|
| MTP off | 12.2 |
MTP on, draft-8 | 17.9 |
MTP on, draft-6 (optimal) | 26.2 |
<sub>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.</sub>
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.

In-class ARC-Challenge comparison (mxfp8), benchmarks by Nightmedia, measured on the base model:
| Model (mxfp8) | ARC-C |
|---|---|
| Qwen3.8-27B-TURBO-Fable (this build's base) | 0.735 |
| Qwen3.6-27B | 0.647 |
| Qwen3.8-27B | 0.591 |
| Qwen3.6-35B-A3B | 0.581 |
| Qwen3.5-27B | 0.557 |
Full seven-benchmark sweep on the base:
| Precision | arc_c | arc_e | boolq | hellaswag | openbookqa | piqa | winogrande |
|---|---|---|---|---|---|---|---|
| mxfp8 | 0.735 | 0.882 | 0.917 | 0.832 | 0.530 | 0.837 | 0.785 |
| mxfp4 | 0.719 | 0.887 | 0.916 | 0.821 | 0.524 | 0.831 | 0.786 |
<sub>Reproduced from the base model card / Nightmedia benchmark runs. These were measured on the base weights, not on this oQ8 repack.</sub>
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.
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.
| Setting | Value | Why |
|---|---|---|
mtp_enabled | true | native MTP accelerates decode (+115%) — real speculative acceptance under oQ8 |
mtp_num_draft_tokens | 6 | swept 6/8 → 6 optimal (26.2 vs 17.9 tok/s) |
qwen35_ane_prefill_enabled | false | net-regresses end-to-end when MTP is active (dispatch overhead) |
turboquant_kv_enabled / turboquant_kv_bits | true / 8 | holds decode up as context grows; tiny KV (hybrid attention) |
qwen35_oq_a8_enabled | false | A8 activations hurt long-context decode in testing |
dflash_enabled | false | net-regressed decode in A/B |
{
"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
}
</details>
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.
# 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?"
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.
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:
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.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}
}
<sub>novæon — digital business architecture + AI · <a href="https://novaeon.studio">novaeon.studio</a></sub>