---
license: mit
pipeline_tag: text-generation
tags:
- cortex.cpp
---
## Overview

**PowerInfer** developed and released the [SmallThinker-3B-preview](https://huggingface.co/PowerInfer/SmallThinker-3B-Preview), a fine-tuned version of the Qwen2.5-3B-Instruct model. SmallThinker is optimized for efficient deployment on resource-constrained devices while maintaining high performance in reasoning, coding, and general text generation tasks. It outperforms its base model on key benchmarks, including AIME24, AMC23, and GAOKAO2024, making it a robust tool for both edge deployment and as a draft model for larger systems like QwQ-32B-Preview.

SmallThinker was fine-tuned in two phases using high-quality datasets, including PowerInfer/QWQ-LONGCOT-500K and PowerInfer/LONGCOT-Refine-500K. Its small size allows for up to 70% faster inference speeds compared to larger models, making it ideal for applications requiring quick responses and efficient computation.

## Variants

| No | Variant | Cortex CLI command |
| --- | --- | --- |
| 1 | [Small-thinker-3b](https://huggingface.co/cortexso/small-thinker/tree/3b) | `cortex run small-thinker:3b` |

## Use it with Jan (UI)

1. Install **Jan** using [Quickstart](https://jan.ai/docs/quickstart)
2. Use in Jan model Hub:
    ```bash
    cortexso/small-thinker
    ```

## Use it with Cortex (CLI)

1. Install **Cortex** using [Quickstart](https://cortex.jan.ai/docs/quickstart)
2. Run the model with command:
    ```bash
    cortex run small-thinker
    ```

## Credits

- **Author:** PowerInfer
- **Converter:** [Homebrew](https://www.homebrew.ltd/)
- **Original License:** [License](https://huggingface.co/PowerInfer/SmallThinker-3B-Preview/blob/main/LICENSE)