# MirrorCode: AI can rebuild entire programs from behavior alone

arXiv: https://arxiv.org/abs/2606.30182 · HTML: https://mlsanity.com/paper/2606.30182/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 32 references that are in the map.
- First posted on arXiv 29 Jun 2026.

## Builds on (most cited first, 17 of 17)

- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code](https://mlsanity.com/paper/2403.07974/) (2024)
- [Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces](https://mlsanity.com/paper/2601.11868/) (2026)
- [SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?](https://mlsanity.com/paper/2509.16941/) (2025)
- [KernelBench: Can LLMs Write Efficient GPU Kernels?](https://mlsanity.com/paper/2502.10517/) (2025)
- [A Careful Examination of Large Language Model Performance on Grade School Arithmetic](https://mlsanity.com/paper/2405.00332/) (2024)
- [RE-Bench: Evaluating frontier AI R&D capabilities of language model agents against human experts](https://mlsanity.com/paper/2411.15114/) (2024)
- [SWE-Lancer: Can Frontier LLMs Earn 1 Million from Real-World Freelance Software Engineering?](https://mlsanity.com/paper/2502.12115/) (2025)
- [Functional Benchmarks for Robust Evaluation of Reasoning Performance, and the Reasoning Gap](https://mlsanity.com/paper/2402.19450/) (2024)
- [ProgramBench: Can Language Models Rebuild Programs From Scratch?](https://mlsanity.com/paper/2605.03546/) (2026)
- [ALE-Bench: A Benchmark for Long-Horizon Objective-Driven Algorithm Engineering](https://mlsanity.com/paper/2506.09050/) (2025)
- [CRUXEval-X: A Benchmark for Multilingual Code Reasoning, Understanding and Execution](https://mlsanity.com/paper/2408.13001/) (2024)
- [SWE-Marathon: Can Agents Autonomously Complete Ultra-Long-Horizon Software Work?](https://mlsanity.com/paper/2606.07682/) (2026)
- [Think Fast: Estimating No-CoT Task-Completion Time Horizons of Frontier AI Models](https://mlsanity.com/paper/2606.07157/) (2026)
- [Vibe Code Bench: Evaluating AI Models on End-to-End Web Application Development](https://mlsanity.com/paper/2603.04601/) (2026)
- [Open-World Evaluations for Measuring Frontier AI Capabilities](https://mlsanity.com/paper/2605.20520/) (2026)
- [A Multi-Language Object-Oriented Programming Benchmark for Large Language Models](https://mlsanity.com/paper/2509.26111/) (2025)

## Built on by (most cited first, 6 of 6)

- [ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks](https://mlsanity.com/paper/2609.18805/) (2026)
- [RECLAIM: Can Agents Reproduce the Claims of Machine Learning Papers?](https://mlsanity.com/paper/2609.28850/) (2026)
- [E2E-SWE: Benchmarking LLMs on Building Working Codebases from Scratch](https://mlsanity.com/paper/2609.38335/) (2026)
- [SWE-Journey: Towards More Realistic Evaluation of Coding Assistants through Long-Horizon, Multi-Turn Interaction](https://mlsanity.com/paper/2610.11559/) (2026)
- [ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction Agents](https://mlsanity.com/paper/2610.13395/) (2026)
- [ReproBAIT: Can coding agents build their own biological AI tools?](https://mlsanity.com/paper/2610.19810/) (2026)

## Abstract (arXiv)

AI models are rapidly improving at autonomous coding, as shown by benchmark progress and one-off demonstrations such as AI implementing a C compiler. However, existing coding benchmarks tend to focus on shorter tasks, and one-off demonstrations are hard to compare systematically because they often have some human guidance, and are not standardized or repeated across models. To address these challenges, we introduce MirrorCode, a long-horizon coding benchmark based on reimplementing entire software projects. In MirrorCode, AI agents must replicate the functionalities of an existing program, without access to its source code. AI solutions must match the original program's output exactly on end-to-end tests, including held-out tests. MirrorCode's 25 target programs span different areas of computing: Unix utilities, data serialization and query tools, bioinformatics, interpreters, static analysis, cryptography, and compression. Existing AI models can already reimplement complex software, with the strongest model scoring 56% across the benchmark. For example, AI can reimplement gotree, a 16,000-line bioinformatics toolkit - a task that we believe would take weeks for a human engineer. However, studying the frontier of performance requires a larger inference budget than typical benchmarks, for example, 2,600 over 19 days for a single attempt on a large task. We show that AI agents can already complete long-horizon software engineering tasks, especially when requirements are precisely specified. More broadly, our work suggests AI will have transformative effects on software engineering, as autonomous agents continue to improve.