Benchmark
SWE-Marathon
SWE-Marathon is an ultra-long-horizon software engineering benchmark covering tasks such as building compilers, optimizing kernels, and developing…
- Modality
- text
- Categories
- agents, code
- Openness
- unknown
- Source
- llm_stats
- Reported scores
- 4
Reported scores
llm_stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| GLM-5.2 | Z.ai | 0.13 | 2026-06-16 | evidence |
| GLM-5.3 | Z.ai | 0.425 | 2026-08-14 | evidence |
| Grok 4.5 | xAI | 0.29 | 2026-07-16 | evidence |
| Kimi K3 | Moonshot AI | 0.42 | 2026-07-16 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.