# SWE-bench

> A benchmark family that evaluates AI systems on real software engineering issues.

- Category: AI-Native Product
- Stage: Established
- Also called: SWE-bench Verified, Software engineering benchmark
- Canonical: https://www.joelcsaji.com/ai-dictionary/swe-bench

## Plain English

It tests whether an AI can fix real code, not just answer coding trivia.

## Example

A model is scored by whether its patch passes tests for a real GitHub issue.

## Why it matters

SWE-bench became a shorthand for practical coding-agent capability.

## Related

- [Coding Agent](https://www.joelcsaji.com/ai-dictionary/coding-agent)
- [Evals](https://www.joelcsaji.com/ai-dictionary/evals)
- [Long-Horizon Task](https://www.joelcsaji.com/ai-dictionary/long-horizon-task)

## Sources

- [Anthropic: Building Effective Agents](https://www.anthropic.com/engineering/building-effective-agents)
- [Stanford HAI: 2026 AI Index](https://hai.stanford.edu/ai-index/2026-ai-index-report)