AI benchmarks and leaderboards are different things
Learn the difference between a benchmark, an evaluation run and a leaderboard, and use each to answer the right question.
Model reports
Agents' Last Exam
Multi-step agentic benchmark using Claude Code harness. Tool Search disabled. Scores depend on harness, reasoning effort, context length, and timeout settings.
Evaluation context and evidence
- Source
- Model reports
- Publisher
- Not recorded
- Benchmark release
- 2025-01-23
- Collected
- Not recorded
- Modality
- Not recorded
These are source-reported observations, not independent reruns by this site. Check the task version, evaluation protocol, unit and reporting date before comparing models. Results from different setups are not a universal ranking.
Original sources
Reported results 2
Shown in source observation order, not ranked. Dates retain the precision provided by the source.
| Model | Value | Unit | Protocol | Reported date and basis | Source |
|---|---|---|---|---|---|
| Hy4 preview | 22.8 | percent | ALE-CLI | 2026-08-28 · document_publication | Hy4 preview |
| GLM-5.3-Flash | 26.3 | percent | official evaluation protocol, Claude Code harness, reasoning effort=max, 1M context, 64K max output, Tool Search disabled, scored by official ALE evaluators | 2026-08-27 · document_publication | GLM-5.3-Flash |