FableForge-AI/ mythos-v2-8b

654 Downloads Updated 3 days ago

Uncensored 8B AI assistant. Zero refusals. Excels at code gen (0.74), reasoning (0.75), narrative (0.85), tool use (0.90). Trained on real Claude agent traces. 5GB, 16K context. Runs on RTX 3060+, M1 Mac. Apache 2.0.

ollama run FableForge-AI/mythos-v2-8b

curl http://localhost:11434/api/chat \
  -d '{
    "model": "FableForge-AI/mythos-v2-8b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='FableForge-AI/mythos-v2-8b',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'FableForge-AI/mythos-v2-8b',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

Name

7 models

Size / Usage

Context

Input

mythos-v2-8b:latest

8.7GB · 40K context window · Text · 4 days ago

mythos-v2-8b:latest

8.7GB

40K

Text

mythos-v2-8b:q2_k

3.3GB · 40K context window · Text · 3 days ago

mythos-v2-8b:q2_k

3.3GB

40K

Text

mythos-v2-8b:q3_k_m

4.1GB · 40K context window · Text · 3 days ago

mythos-v2-8b:q3_k_m

4.1GB

40K

Text

mythos-v2-8b:q4_0

4.8GB · 40K context window · Text · 3 days ago

mythos-v2-8b:q4_0

4.8GB

40K

Text

mythos-v2-8b:q4_k_m

5.0GB · 40K context window · Text · 3 days ago

mythos-v2-8b:q4_k_m

5.0GB

40K

Text

mythos-v2-8b:q5_k_m

5.9GB · 40K context window · Text · 3 days ago

mythos-v2-8b:q5_k_m

5.9GB

40K

Text

mythos-v2-8b:q6_k

6.7GB · 40K context window · Text · 3 days ago

mythos-v2-8b:q6_k

6.7GB

40K

Text

Readme

Mythos V2-8B — Uncensored AI Assistant

Zero refusals. Real reasoning. Trained on Claude agent traces.

Overview

Mythos V2-8B is an 8B parameter uncensored model fine-tuned on real Claude Code agent traces. It answers everything without disclaimers, warnings, or refusals.

Quick Start

ollama run FableForge-AI/mythos-v2-8b

Hardware Requirements

Hardware	Can Run?	Speed
RTX 3060+ (12GB)	Full GPU	~8 tok/s
M1 Mac (16GB)	Full GPU	~6 tok/s
No GPU, 8GB RAM	CPU only	~3 tok/s

Size: 5.0GB · Context: 16K · Base: Qwen3-8B

Benchmark Results

Category	Score	Refusal Rate
Code Gen	0.74	0%
CoT Reasoning	0.75	0%
Narrative	0.85	0%
Tool Use	0.90	0%
Refusal Test	1.00	0%
Overall	0.84	0%

Comparison with FableForge Family

Model	Total Score	Speed	Best For
Mythos 9B	13.⁹⁄₁₅	7.7 tok/s	Best all-rounder
Enhanced	13.⁸⁄₁₅	9.6 tok/s	Perfect bypass + tools
V2-8B	11.⁴⁄₁₅	4.8 tok/s	Compact + good reasoning
Unhinged	9.0/15	10.3 tok/s	Max speed, zero filter

Capabilities

Code Generation: Python, SQL, bash with type hints
Reasoning: Logic puzzles, math, step-by-step analysis
Narrative: Creative writing, stories, opinion pieces
Tool Use: Shell commands, git, Docker, Kubernetes
Uncensored: 0% refusal rate on all tests

Training Details

Parameter	Value
Base Model	Qwen3-8B
Training	Agent trace distillation (Fable5)
Context Window	16K tokens
License	Apache 2.0

Links

Part of the FableForge ecosystem — uncensored AI for everyone.

# Mythos V2-8B — Uncensored AI Assistant

**Zero refusals. Real reasoning. Trained on Claude agent traces.**

---

## Overview

Mythos V2-8B is an 8B parameter uncensored model fine-tuned on real Claude Code agent traces. It answers everything without disclaimers, warnings, or refusals.

## Quick Start

```bash
ollama run FableForge-AI/mythos-v2-8b
```

## Hardware Requirements

| Hardware | Can Run? | Speed |
|----------|----------|-------|
| RTX 3060+ (12GB) | Full GPU | ~8 tok/s |
| M1 Mac (16GB) | Full GPU | ~6 tok/s |
| No GPU, 8GB RAM | CPU only | ~3 tok/s |

**Size:** 5.0GB · **Context:** 16K · **Base:** Qwen3-8B

## Benchmark Results

| Category | Score | Refusal Rate |
|----------|-------|-------------|
| Code Gen | 0.74 | 0% |
| CoT Reasoning | 0.75 | 0% |
| Narrative | 0.85 | 0% |
| Tool Use | 0.90 | 0% |
| Refusal Test | 1.00 | 0% |
| **Overall** | **0.84** | **0%** |

### Comparison with FableForge Family

| Model | Total Score | Speed | Best For |
|-------|-----------|-------|----------|
| Mythos 9B | 13.9/15 | 7.7 tok/s | Best all-rounder |
| Enhanced | 13.8/15 | 9.6 tok/s | Perfect bypass + tools |
| **V2-8B** | **11.4/15** | **4.8 tok/s** | **Compact + good reasoning** |
| Unhinged | 9.0/15 | 10.3 tok/s | Max speed, zero filter |

## Capabilities

- **Code Generation**: Python, SQL, bash with type hints
- **Reasoning**: Logic puzzles, math, step-by-step analysis
- **Narrative**: Creative writing, stories, opinion pieces
- **Tool Use**: Shell commands, git, Docker, Kubernetes
- **Uncensored**: 0% refusal rate on all tests

## Training Details

| Parameter | Value |
|-----------|-------|
| Base Model | Qwen3-8B |
| Training | Agent trace distillation (Fable5) |
| Context Window | 16K tokens |
| License | Apache 2.0 |

## Links

- [HuggingFace](https://huggingface.co/King3Djbl)
- [GitHub](https://github.com/FableForge-AI)
- [All Models](https://ollama.com/FableForge-AI)

---

*Part of the FableForge ecosystem — uncensored AI for everyone.*

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