bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning

bazobehram/ qwen3-14b-claude-4.5-opus-high-reasoning

1,542 Downloads Updated 1 month ago

ollama run bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning

curl http://localhost:11434/api/chat \
  -d '{
    "model": "bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

Name

1 model

Size

Context

Input

qwen3-14b-claude-4.5-opus-high-reasoning:latest

9.0GB · 40K context window · Text · 1 month ago

qwen3-14b-claude-4.5-opus-high-reasoning:latest

9.0GB

40K

Text

Readme

# Qwen3-14B Claude 4.5 Opus High-Reasoning (GGUF)

Qwen3-14B-Claude-4.5-Opus-High-Reasoning is a high-performance 14.8B parameter model optimized for deep logical inference and complex problem-solving. It is a fine-tuned version of the Qwen3-14B base model, utilizing a specialized dataset distilled from Claude 4.5 Opus to replicate its sophisticated step-by-step reasoning patterns.

## 🚀 Key Features

* **Distilled Reasoning:** Inherits the high-level logic, planning, and multi-step derivation capabilities of Claude 4.5 Opus.
* **Chain-of-Thought (CoT):** Explicitly trained to use `<thought>` blocks to "think" through problems internally before delivering a final answer, significantly reducing hallucinations.
* **Large Context Window:** Supports up to 40,000 tokens, making it ideal for deep document analysis and long-form coding projects.
* **Optimized Quantization:** Provided in Q4_K_M (9.0GB), offering a perfect balance between intelligence and speed for local deployment on consumer GPUs.

## 🧠 How to Use

For the best results, this model requires a system prompt that encourages its internal reasoning process.

### Quick Start

```bash
ollama run bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning

Recommended System Prompt

You can set this in your Modelfile or pass it during API calls:

“You are a helpful assistant. For every query, provide a detailed, step-by-step reasoning process within the thought block before giving the final answer.”

📊 Model Details

Parameters: 14.8 Billion
Context Length: 40,960 tokens
Quantization: Q4_K_M (GGUF)
Architecture: Qwen2.5/Qwen3
Primary Use Cases: Advanced coding, mathematical proofs, scientific analysis, and logical derivation.

💻 Hardware Requirements

VRAM: 12GB+ (Recommended) or 8GB (Minimum with context offloading).
System RAM: 16GB+
GPU: Works best on NVIDIA RTX ³⁰⁶⁰⁄₄₀₇₀ (12GB) or Apple M-series (16GB+ Unified Memory).

⚖️ Attribution & Credits

Original Model & Dataset: Developed by TeichAI (Hugging Face Repository)
Base Model: unsloth/Qwen3-14B
Distillation Source: Claude 4.5 Opus (Anthropic)
Ported to Ollama by: bazobehram

```markdown
# Qwen3-14B Claude 4.5 Opus High-Reasoning (GGUF)

Qwen3-14B-Claude-4.5-Opus-High-Reasoning is a high-performance 14.8B parameter model optimized for deep logical inference and complex problem-solving. It is a fine-tuned version of the Qwen3-14B base model, utilizing a specialized dataset distilled from Claude 4.5 Opus to replicate its sophisticated step-by-step reasoning patterns.

## 🚀 Key Features

* **Distilled Reasoning:** Inherits the high-level logic, planning, and multi-step derivation capabilities of Claude 4.5 Opus.
* **Chain-of-Thought (CoT):** Explicitly trained to use `<thought>` blocks to "think" through problems internally before delivering a final answer, significantly reducing hallucinations.
* **Large Context Window:** Supports up to 40,000 tokens, making it ideal for deep document analysis and long-form coding projects.
* **Optimized Quantization:** Provided in Q4_K_M (9.0GB), offering a perfect balance between intelligence and speed for local deployment on consumer GPUs.

## 🧠 How to Use

For the best results, this model requires a system prompt that encourages its internal reasoning process.

### Quick Start

```bash
ollama run bazobehram/qwen3-14b-claude-4.5-opus-high-reasoning

```

### Recommended System Prompt

You can set this in your `Modelfile` or pass it during API calls:

> "You are a helpful assistant. For every query, provide a detailed, step-by-step reasoning process within the thought block before giving the final answer."

## 📊 Model Details

* **Parameters:** 14.8 Billion
* **Context Length:** 40,960 tokens
* **Quantization:** Q4_K_M (GGUF)
* **Architecture:** Qwen2.5/Qwen3
* **Primary Use Cases:** Advanced coding, mathematical proofs, scientific analysis, and logical derivation.

## 💻 Hardware Requirements

* **VRAM:** 12GB+ (Recommended) or 8GB (Minimum with context offloading).
* **System RAM:** 16GB+
* **GPU:** Works best on NVIDIA RTX 3060/4070 (12GB) or Apple M-series (16GB+ Unified Memory).

## ⚖️ Attribution & Credits

* **Original Model & Dataset:** Developed by TeichAI ([Hugging Face Repository](https://huggingface.co/TeichAI/Qwen3-14B-Claude-4.5-Opus-High-Reasoning-Distill-GGUF))
* **Base Model:** unsloth/Qwen3-14B
* **Distillation Source:** Claude 4.5 Opus (Anthropic)
* **Ported to Ollama by:** bazobehram

```
```

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