fredrezones55/qwen3.5-opus:27b

fredrezones55/ qwen3.5-opus:27b

2,028 Downloads Updated 4 days ago

Qwen3.5-Claude-4.6-Opus-Reasoning-Distilled-v2; https://huggingface.co/Jackrong/; has vision properly merged and efficiently quantified.

vision tools thinking 4b 9b 27b

ollama run fredrezones55/qwen3.5-opus:27b

curl http://localhost:11434/api/chat \
  -d '{
    "model": "fredrezones55/qwen3.5-opus:27b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='fredrezones55/qwen3.5-opus:27b',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'fredrezones55/qwen3.5-opus:27b',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 4 days ago

4 days ago

914789b72c24 · 17GB ·

model

archqwen35

·

parameters27.4B

·

quantizationQ4_K_M

17GB

params

{ "presence_penalty": 1.5, "temperature": 1, "top_k": 20, "top_p": 0.95 }

65B

template

{{ .Prompt }}

13B

Readme

Qwen3.5 is a thinking, vision, and tooling base model; that means Qwen3.5 fine-tunes for Ollama should also have these traits. Until now! Taken from R&D of work porting qwen3-VL Jan fine-tunes (had to fix some issues in the pipeline) these qwen3.5 models have had gone through a transformation to properly handle Ollama’s “New Engine”’s quirks. [I have not, however noticed a structure issue {thinking breaks} and that would be more of a configuration problem] {I do not have 64GBs of RAM (in this economy!?) to potentially handle HF to Ollama via ‘Ollama create’, so I had to get handsy with the model blobs}

Need to get a better config for the parameters, I’m not sure if the defaults are right…

Models:

4B

9B - for general use

9B-tooling - for coding and precise tooling

27B

Log: [3/21/2026] - I’m shooked, there is a 27B model to also work on and merge. [fun…]

[3/25/2026] - 27B model added, will wait for the 30b MOE model.

Model Information and sources 27b model 9b model 4b model

Qwen3.5 is a **thinking, vision, and tooling base model**; that means Qwen3.5 *fine-tunes for Ollama* should also have these traits. Until now! Taken from R&D of work porting qwen3-VL Jan fine-tunes (had to fix some issues in the pipeline) these qwen3.5 models have had gone through a transformation to properly handle Ollama's "New Engine"'s quirks. [I have not, however noticed a structure issue {*thinking breaks*} and that would be more of a configuration problem] {I do not have 64GBs of RAM (*in this economy!?*) to potentially handle HF to Ollama via 'Ollama create', so I had to get handsy with the model blobs}

---
Need to get a better config for the parameters, I'm not sure if the defaults are right...

Models:

4B

9B - for general use

9B-tooling - for coding and precise tooling

27B

---
Log:
[3/21/2026] - I'm shooked, there is a 27B model to also work on and merge. [fun...]

[3/25/2026] - 27B model added, will wait for the 30b MOE model.

---

Model Information and sources
[27b model](https://huggingface.co/Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-GGUF)
[9b model](https://huggingface.co/Jackrong/Qwen3.5-9B-Claude-4.6-Opus-Reasoning-Distilled-v2-GGUF)
[4b model](https://huggingface.co/Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-v2-GGUF)

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)