---
title: "Empero's Qwen3.8-2B Doubles Reasoning Scores Running on a Phone"
subtitle: "Empero's tiny reasoning model squeezes a 2.4 trillion parameter teacher into a 1.3 GB file that runs on phones with a 262k context window."
url: "https://alphasignal.ai/news/empero-s-qwen3-8-2b-doubles-reasoning-scores-running-on-a-phone"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-08-15T18:08:46.000Z"
updated: "2026-08-27T19:55:08.291Z"
type: "Model"
topics:
  - "Llms"
  - "Gpus"
subtopics:
  - "Quantization"
  - "Edge Deployment"
upvotes: 328
source_url: "https://huggingface.co/empero-ai/Qwen3.8-2B-Distill-GGUF"
comments: 0
truncated: true
---

# Empero's Qwen3.8-2B Doubles Reasoning Scores Running on a Phone

> Empero's tiny reasoning model squeezes a 2.4 trillion parameter teacher into a 1.3 GB file that runs on phones with a 262k context window.

Empero has quietly shipped GGUF quantizations of [Qwen3.8-2B-Distill](https://huggingface.co/empero-ai/Qwen3.8-2B-Distill), a 2 billion parameter reasoning model distilled from a much larger mixture-of-experts teacher. The recommended Q4\_K\_M file weighs 1.31 GB, small enough to run on a phone, yet it nearly doubles its own base model's MMLU score. Downloads on Hugging Face have crossed 171,000.

## A 2.4T teacher, squeezed into 2B

The release consists of [GGUF quantizations](https://huggingface.co/empero-ai/Qwen3.8-2B-Distill-GGUF) of empero-ai/Qwen3.8-2B, a full-parameter distillation of Qwen3.8 2.4T A95B into the Qwen3.5-2B architecture, the smallest member of the family. The GGUFs target llama.cpp, Ollama, LM Studio, Jan, KoboldCpp, and other stock runtimes, so the reasoning behavior of a frontier-scale sparse teacher now runs inside a dense 2B student on almost anything.

The training recipe is straightforward supervised fine-tuning on teacher traces. According to the model card, the student saw roughly 30,000 curated teacher traces from Empero's internal Qwen3.8 distillation datasets. For the sibling 9B model, Empero has stated the process uses off-policy supervised fine-tuning rather than reinforcement learning, which keeps the pipeline simple and reproducible.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [empero-ai/Qwen3.8-2B-Distill-GGUF on Hugging Face](https://huggingface.co/empero-ai/Qwen3.8-2B-Distill-GGUF)
- [empero-ai/Qwen3.8-2B-Distill on Hugging Face](https://huggingface.co/empero-ai/Qwen3.8-2B-Distill)
- [Qwen3.8-9B Distill by Empero: Benchmarks and Local Setup](https://www.mindstudio.ai/blog/qwen3-8-9b-distill-empero)
- [empero-ai/Qwen3.8-9B-Distill on Hugging Face](https://huggingface.co/empero-ai/Qwen3.8-9B-Distill)