Benchmark
Agents' Last Exam
Multi-step agentic benchmark using Claude Code harness. Tool Search disabled. Scores depend on harness, reasoning effort, context length, and timeout…
- Released
- 2025-01-23
- Categories
- coding_agent
- Openness
- unknown
- Source
- model_reports
- Reported scores
- 2
Reported scores
model_reports
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| GLM-5.3-Flash | Z.ai | 26.3 | 2026-08-27 | evidence |
| Hy4 preview | Tencent | 22.8 | 2026-08-28 | 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.