# Satori-SWE: Evolutionary Test-Time Scaling for Sample-Efficient Software Engineering

arXiv: https://arxiv.org/abs/2505.23604 · HTML: https://mlsanity.com/paper/2505.23604/

## Key facts

- Cited by 17 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 20 of its 37 references that are in the map.
- First posted on arXiv 29 May 2025.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)

## Builds on (most cited first, 20 of 20)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Training Compute-Optimal Large Language Models](https://mlsanity.com/paper/2203.15556/) (2022)
- [Qwen2.5-Coder Technical Report](https://mlsanity.com/paper/2409.12186/) (2024)
- [Large Language Monkeys: Scaling Inference Compute with Repeated Sampling](https://mlsanity.com/paper/2407.21787/) (2024)
- [DeepSeek-V3 Technical Report](https://mlsanity.com/paper/2412.19437/) (2024)
- [Evolution Strategies as a Scalable Alternative to Reinforcement Learning](https://mlsanity.com/paper/1703.03864/) (2017)
- [Agentless: Demystifying LLM-based Software Engineering Agents](https://mlsanity.com/paper/2407.01489/) (2024)
- [The CMA Evolution Strategy: A Tutorial](https://mlsanity.com/paper/1604.00772/) (2016)
- [SWE-smith: Scaling Data for Software Engineering Agents](https://mlsanity.com/paper/2504.21798/) (2025)
- [SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution](https://mlsanity.com/paper/2502.18449/) (2025)
- [REINFORCE++: Stabilizing Critic-Free Policy Optimization with Global Advantage Normalization](https://mlsanity.com/paper/2501.03262/) (2025)
- [R2E-Gym: Procedural Environments and Hybrid Verifiers for Scaling Open-Weights SWE Agents](https://mlsanity.com/paper/2504.07164/) (2025)
- [SWE-Fixer: Training Open-Source LLMs for Effective and Efficient GitHub Issue Resolution](https://mlsanity.com/paper/2501.05040/) (2025)
- [OpenRLHF: An Easy-to-use, Scalable and High-performance RLHF Framework](https://mlsanity.com/paper/2405.11143/) (2024)
- [Lingma SWE-GPT: An Open Development-Process-Centric Language Model for Automated Software Improvement](https://mlsanity.com/paper/2411.00622/) (2024)
- [Satori: Reinforcement Learning with Chain-of-Action-Thought Enhances LLM Reasoning via Autoregressive Search](https://mlsanity.com/paper/2502.02508/) (2025)
- [CodeMonkeys: Scaling Test-Time Compute for Software Engineering](https://mlsanity.com/paper/2501.14723/) (2025)
- [Thinking Longer, Not Larger: Enhancing Software Engineering Agents via Scaling Test-Time Compute](https://mlsanity.com/paper/2503.23803/) (2025)

## Built on by (most cited first, 17 of 17)

- [The Landscape of Agentic Reinforcement Learning for LLMs: A Survey](https://mlsanity.com/paper/2509.02547/) (2025)
- [A Survey of Reinforcement Learning for Large Reasoning Models](https://mlsanity.com/paper/2509.08827/) (2025)
- [From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence](https://mlsanity.com/paper/2511.18538/) (2025)
- [SWE-Replay: Efficient Test-Time Scaling for Software Engineering Agents](https://mlsanity.com/paper/2601.22129/) (2026)
- [Confucius Code Agent: Scalable Agent Scaffolding for Real-World Codebases](https://mlsanity.com/paper/2512.10398/) (2025)
- [Beyond Pipelines: A Survey of the Paradigm Shift toward Model-Native Agentic AI](https://mlsanity.com/paper/2510.16720/) (2025)
- [Agentic Software Issue Resolution with Large Language Models: A Survey](https://mlsanity.com/paper/2512.22256/) (2025)
- [Hindsight Hint Distillation: Scaffolded Reasoning for SWE Agents from CoT-free Answers](https://mlsanity.com/paper/2605.11556/) (2026)
- [Tailored Primitive Initialization is the Secret Key to Reinforcement Learning](https://mlsanity.com/paper/2511.12429/) (2025)
- [Independent Patch Verification for Coding Agents with a Bidirectional Reconstruct-and-Verify Framework](https://mlsanity.com/paper/2608.08950/) (2026)
- [EGSS: Entropy-guided Stepwise Scaling for Reliable Software Engineering](https://mlsanity.com/paper/2602.05242/) (2026)
- [Pushing Forward Pareto Frontiers of Proactive Agents with Behavioral Agentic Optimization](https://mlsanity.com/paper/2602.11351/) (2026)
- [Decocted Experience Improves Test-Time Inference in LLM Agents](https://mlsanity.com/paper/2604.04373/) (2026)
- [Learning Generalizable Behaviors for Terminal Agents](https://mlsanity.com/paper/2608.22631/) (2026)
- [Rethinking Agentic Reinforcement Learning In Large Language Models](https://mlsanity.com/paper/2604.27859/) (2026)
- [FLARE: A Full-Lifecycle Dense Supervision Paradigm for Long-Horizon Coding Agents via Generative Reward Model](https://mlsanity.com/paper/2609.23808/) (2026)
- [FC-SWE: Failure-Conditioned RL for Long-Horizon Software Engineering Agents](https://mlsanity.com/paper/2610.07898/) (2026)

## Abstract (arXiv)

Language models (LMs) perform well on standardized coding benchmarks but struggle with real-world software engineering tasks such as resolving GitHub issues in SWE-Bench, especially when model parameters are less than 100B. While smaller models are preferable in practice due to their lower computational cost, improving their performance remains challenging. Existing approaches primarily rely on supervised fine-tuning (SFT) with high-quality data, which is expensive to curate at scale. An alternative is test-time scaling: generating multiple outputs, scoring them using a verifier, and selecting the best one. Although effective, this strategy often requires excessive sampling and costly scoring, limiting its practical application. We propose Evolutionary Test-Time Scaling (EvoScale), a sample-efficient method that treats generation as an evolutionary process. By iteratively refining outputs via selection and mutation, EvoScale shifts the output distribution toward higher-scoring regions, reducing the number of samples needed to find correct solutions. To reduce the overhead from repeatedly sampling and selection, we train the model to self-evolve using reinforcement learning (RL). Rather than relying on external verifiers at inference time, the model learns to self-improve the scores of its own generations across iterations. Evaluated on SWE-Bench-Verified, EvoScale enables our 32B model, Satori-SWE-32B, to match or exceed the performance of models with over 100B parameters while using a few samples. Code, data, and models will be fully open-sourced.