Qwen3.8-27B-JANG_4D

OsaurusAI/Qwen3.8-27B-JANG_4D

Qwen3.8-27B-JANG_4D is a image-text-to-text model from OsaurusAI. 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.

<p align="center" <a href="https://osaurus.ai"<img src="./osaurus-x-banner.png" alt="Osaurus AI"</a </p

Downloads · 30 days

319

21% of all-time downloads

All-time downloads

1.5K

Public

Parameters

27.4B

75.4 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors26.5 GB · 100%

Parameter types

How the weights are stored.

U3227.2B · 99%

Try a prompt

Base models

Qwen/Qwen3.8-27B

Model card

From the Hugging Face model README

Runtime: Osaurus 0.25.20 or newer (osaurus.json: required_osaurus_version 0.25.20, model_version 2).

OsaurusAI/Qwen3.8-27B-JANG_4D

Iteration 2 (2026-10-05): weights replaced. New calibration (text + image/video + own bf16 traces at every effort), every attention-type layer at 8-bit, MTP head removed in favour of a bundled DFlash2 drafter. Re-download if you have the earlier version.

Qwen3.8-27B at near-reference quality in 21 GiB, with image and video input and a bundled DFlash2 drafter.

part	format
MLP gate / up / down	affine 4 / 5 / 6 / 8-bit, picked per projection by measured error per byte
full attention (q / k / v / o) and gated-delta-net projections	affine 8-bit
`lm_head`, token embeddings, vision tower	affine 8-bit
norms, convolutions, gates	bf16

All affine modules: importance-matrix weighted fit, AWQ folded into the producing norm, GPTQ on the MLX grid below 8 bits.

Quality vs the bf16 model

set	median KL	top-1
mixed held-out text	0.0007	95.2 %
agentic and tool conversations	0.0014	92.1 %
reasoning	0.0001	96.9 %
own bf16 reasoning traces, effort low	0.0001	93.9 %
own bf16 reasoning traces, effort medium	0.00004	97.7 %
own bf16 reasoning traces, effort xhigh	0.0005	95.1 %

Think-block close decisions agree with bf16 on 40 / 40 held-out traces and 65 / 65 text rows; tool-call vs answer decisions flip on 9 / 263 agentic positions.

Quality by topic

Median KL and top-1 agreement vs the bf16 model per topic of the held-out text set (both bundles, same rows):

topic	tokens scored	JANG_4D KL	JANG_4D top-1	JANGH2 KL	JANGH2 top-1
coding (many languages)	39,155	0.0002	98.4 %	0.0092	89.5 %
math	9,067	0.0001	98.7 %	0.0068	90.8 %
office / finance	13,159	0.0001	97.7 %	0.0087	86.9 %
reasoning	14,004	0.0001	94.5 %	0.0081	81.0 %
science	4,156	0.0010	98.0 %	0.0460	87.1 %
long context + attention recall	79,784	0.0007	96.6 %	0.0446	80.4 %
cybersecurity	15,683	0.0008	94.9 %	0.0368	80.5 %
networking	6,458	0.0011	93.7 %	0.0244	81.9 %
terminal / shell	42,987	0.0013	92.5 %	0.0738	73.5 %
agentic + tool calls	34,245	0.0026	89.8 %	0.0975	71.2 %
creative writing	10,029	0.0028	96.5 %	0.1989	74.2 %
multi-turn agentic set	199,746	0.0014	92.1 %	0.0615	75.3 %

Own bf16 reasoning traces, by topic (thinking + answer tokens):

topic	JANG_4D KL	JANG_4D top-1	JANGH2 KL	JANGH2 top-1
coding (many languages)	0.00007	98.4 %	0.00508	89.4 %
math	0.00001	98.6 %	0.00061	91.6 %
office / finance	0.00000	98.5 %	0.00007	91.2 %
science	0.00010	98.1 %	0.00779	87.8 %
networking	0.00000	98.7 %	0.00009	92.8 %
cybersecurity	0.00135	97.3 %	0.08516	81.1 %
agentic + tool calls	0.00093	89.7 %	0.03968	72.6 %

Reasoning, tools, sampling

Thinking is on by default (the template opens <think>); enable_thinking: false prefills a closed block.
reasoning_effort: low, medium, xhigh (default); the template rejects anything else. Earlier thinking is kept in history by default (preserve_thinking).
Reasoning parser qwen3; tool calls use the XML function format (parser qwen3_coder).
Sampling presets from the model card: thinking T 1.0 / top_p 0.95 / top_k 20; instruct T 0.7 / top_p 0.8 / top_k 20 / presence 1.5.

DFlash2 drafter

dflash2/ holds the z-lab/Qwen3.8-27B-DFlash2 block-diffusion drafter (bf16, block size 8) for speculative decoding of text; image and video requests decode without it. The native MTP head is not included.

Calibration

One capture feeds the importance matrix, AWQ and GPTQ, merged by token count: 1.21M text tokens (cybersecurity, coding in many languages, agentic and tool conversations, terminal, long reasoning, attention recall, math; all three efforts), 0.43M image and video tokens (191 media items through the vision tower), and 0.31M tokens of the model's own full-precision reasoning traces at low / medium / xhigh. Held-out sets are disjoint from calibration.

What's in the bundle

model-*.safetensors + one index (64-byte aligned), config.json (per-module quantization), jang_config.json (format, capabilities, reasoning and tool contract, sampling, evaluation), generation_config.json, tokenizer, chat template, image and video processor configs, dflash2/.

Quantized and validated by Jinho Jang — [email protected]

Task: Image-Text-to-Text
Library: mlx
Type: qwen3_5
License: apache-2.0
Languages: en
Created: Aug 14, 2026
Updated: Oct 7, 2026