> ## Documentation Index
> Fetch the complete documentation index at: https://rllm-org-rllm-19-feat-renderer-parser-backend.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# DeepCoder

> A fully open-source 14B coder matching O3-mini on competitive programming

<Card title="Read the full write-up" icon="arrow-up-right-from-square" href="https://pretty-radio-b75.notion.site/DeepCoder-A-Fully-Open-Source-14B-Coder-at-O3-mini-Level-1cf81902c14680b3bee5eb349a512a51">
  Notion blog post with full details
</Card>

DeepCoder is a 14B coding model that achieves **60.6% Pass\@1 on LiveCodeBench**, an 8% improvement over the base model. It matches `o3-mini` and `o1` performance on competitive programming benchmarks.

## Results

| Model         | Parameters | LiveCodeBench Pass\@1 |
| ------------- | ---------- | --------------------- |
| **DeepCoder** | **14B**    | **60.6%**             |
| o3-mini (Low) | Unknown    | \~60%                 |
| o1            | Unknown    | \~60%                 |

## Approach

Fine-tuned from DeepSeek-R1-Distill-Qwen-14B on coding competition problems using RL with code execution feedback.

See the [`cookbooks/deepcoder`](https://github.com/rllm-org/rllm/tree/main/cookbooks/deepcoder) cookbook (single-turn coding with hidden-test grading) for the AgentFlow-based reproducer.

**Released**: April 2025