# Qwythos-9B-Claude-Mythos-5-1M

Qwythos-9B is a full-parameter reasoning model built on top of a deeply uncensored Qwen3.5-9B base and post-trained on over 500 million tokens of high-quality Claude Mythos and Claude Fable traces, with chain-of-thought generated in-house by Empero AI's internal tool rethink.

- **Kind:** model (Hugging Face)
- **Hugging Face:** https://huggingface.co/empero-ai/Qwythos-9B-Claude-Mythos-5-1M
- **Owner:** empero-ai (https://forefy.com/u/hf/empero-ai)
- **Parameters:** 9.4B
- **Context:** 1048576 tokens
- **License:** apache-2.0
- **Focus:** general
- **Architecture:** Qwen3_5ForConditionalGeneration
- **Pinned revision:** `14a29bae5143091aeaf87ad37120de4cd57d592c`
- **Page:** https://forefy.com/models/a7f72918-b502-5934-a7a3-68f572cab39c
- **API (JSON):** https://forefy.com/api/asr/models/a7f72918-b502-5934-a7a3-68f572cab39c

## What was done to it

- uncensored (declared evidence)
- finetune (declared evidence, from Qwen/Qwen3.5-9B)

## Lineage

Qwen/Qwen3.5-9B-Base → Qwen/Qwen3.5-9B → empero-ai/Qwythos-9B-Claude-Mythos-5-1M

## Checks (repo at the pinned revision; not a behavior test)

- **Identity** [pass]: No lookalike name or conflicting provenance
- **Weights format** [pass]: safetensors only, nothing runs on load
- **Pickle files** [pass]: None.
- **Remote code** [pass]: None
- **Chat template** [warn]: Differs from Qwen/Qwen3.5-9B's
- **Refusals modified** [info]: Yes: uncensoring (declared)
- **License** [pass]: apache-2.0
- **Lineage** [pass]: Derives from Qwen/Qwen3.5-9B
- **Model card** [pass]: 5 of 6, missing evaluation results
- **Safetensors headers** [pass]: Every byte is indexed, no hidden data
- **Tensor layout** [warn]: 0 tensor(s) added, 15 removed vs Qwen/Qwen3.5-9B

## Variants

- BF16 safetensors (safetensors, 18.8 GB)

## Run it

- transformers: `AutoModelForCausalLM.from_pretrained("empero-ai/Qwythos-9B-Claude-Mythos-5-1M", revision="14a29bae5143091aeaf87ad37120de4cd57d592c")`