---

base_model: TeichAI/Qwen3.8-27B-Fable-Distill

base_model_relation: quantized

library_name: gguf

pipeline_tag: image-text-to-text

license: apache-2.0

language:

• en

tags:

• gguf
• llama.cpp
• qwen3_5
• quantized
• vision

datasets:

• armand0e/claude-fable-5-claude-code
• armand0e/Fable-5-Chat

---

Qwen3.8-27B-Fable-Distill — GGUF

!Benchmark Comparison

| Model | ARC Challenge | ARC Challenge (Easy) | BoolQ |

|---|---|---|---|

| Qwen3.8-27B | 0.591 | 0.782 | 0.896 |

| Qwen3.8-27B-Fable-Distill | 0.637 | 0.832 | 0.911 |

As always, big thank you to @nightmedia for the benchmarks

GGUF conversions of TeichAI/Qwen3.8-27B-Fable-Distill,

a BF16 finetune of Qwen3.8-27B (base: Qwen/Qwen3.8-27B) trained with Unsloth + TRL.

The model was trained on a public set of chat and agent traces from Fable 5 as well as a much larger corpus of private personal Fable 5 data.

Converted with llama.cpp b6b4344e.

MTP head kept at BF16

This model ships a multi-token-prediction (nextn) head, and **every quant here

keeps that head unquantized at BF16** while the other 64 layers are quantized

normally:

qwen35.block_count          = 65      # 64 transformer layers + 1 MTP layer
qwen35.nextn_predict_layers = 1
blk.64.*                    = bf16    # 424.7M params, left alone

Files

| File | Bits | Size | Notes |

|---|---|---|---|

| BF16/…-BF16-*.gguf | 16 | ~55 GB | Full precision, split into shards. Convert your own quants from this. |

| …-Q8_0.gguf | 8 | ~29 GB | Near-lossless. |

| …-Q6_K.gguf | 6 | ~23 GB | Very close to Q8_0 at meaningfully smaller size. |

| …-Q5_K_M.gguf | 5 | ~20 GB | Strong quality/size balance. |

| …-Q5_K_S.gguf | 5 | ~19 GB | |

| …-Q4_K_M.gguf | 4 | ~17 GB | Recommended default for most users. |

| …-Q4_K_S.gguf | 4 | ~16 GB | Slightly smaller than Q4_K_M. |

| …-IQ4_NL.gguf | 4 | ~16 GB | Non-linear 4-bit. |

| …-IQ4_XS.gguf | 4 | ~16 GB | Smallest of the 4-bit family. |

| …-Q3_K_L.gguf | 3 | ~15 GB | |

| …-Q3_K_M.gguf | 3 | ~14 GB | |

| …-Q3_K_S.gguf | 3 | ~13 GB | |

| …-Q2_K.gguf | 2 | ~11 GB | Noticeable quality loss. |

| mmproj-F32.gguf | 32 | 1.8 GB | Vision projector, full precision. |

| mmproj-BF16.gguf | 16 | 0.9 GB | Vision projector, bfloat16. |

| mmproj-F16.gguf | 16 | 0.9 GB | Vision projector, float16. Fine for nearly everyone. |

Usage

Text:

llama-cli -m Qwen3.8-27B-Fable-Distill-Q4_K_M.gguf -c 8192 -p "Hello"

Vision — pass the projector alongside the model:

llama-mtmd-cli -m Qwen3.8-27B-Fable-Distill-Q4_K_M.gguf \
               --mmproj mmproj-F16.gguf \
               --image photo.jpg -p "Describe this image."

Server:

llama-server -m Qwen3.8-27B-Fable-Distill-Q4_K_M.gguf --mmproj mmproj-F16.gguf

Server + MTP:

llama-server -m Qwen3.8-27B-Fable-Distill-Q4_K_M.gguf --mmproj mmproj-F16.gguf --spec-type draft-mtp --spec-draft-n-max 3

Notes

• The model is multimodal (image-text-to-text). Without an mmproj-*.gguf you

get a text-only model. Three precisions are provided; F16 is the usual

choice, BF16 matches the source weights' dtype, and F32 is there if you

want the projector left entirely unquantized.

• Qwen3.5-family chat template with thinking support: it accepts

enable_thinking and a reasoning_effort of low, medium or xhigh

(the template's own default is xhigh, which thinks at length every turn).

• Base model sampling recommendations: temperature 1.0, top_p 0.95, top_k 20.

---

The data for this model was easily formatted, validated, and masked using Teich <img src="https://cdn-avatars.huggingface.co/v1/production/uploads/6837935ac3b7ffe0d2559ce9/-AxyvV4wfUY8uo87kNKkK.png" width="20" height="20" style="display: inline-block; vertical-align: middle; margin: 0 3px;">

This qwen3_5 model was trained 2x faster with Unsloth and Huggingface's TRL library.