Benchmark
Agents' Last Exam
Agents' Last Exam is a challenging benchmark for AI agents on hard, long-horizon tasks that test sustained reasoning, planning, and tool use, reported…
- Modality
- text
- Categories
- reasoning, agents, tool_calling
- Openness
- unknown
- Source
- llm_stats
- Reported scores
- 10
Reported scores
llm_stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| DeepSeek-V4-Flash-0731 | DeepSeek | 0.252 | 2026-07-31 | evidence |
| DeepSeek-V4-Pro-0813 | DeepSeek | 0.257 | 2026-08-13 | evidence |
| Gemini 3.7 Flash | 0.263 | 2026-08-13 | evidence | |
| GLM-5.3 | Z.ai | 0.285 | 2026-08-14 | evidence |
| GPT-5.6 Luna | OpenAI | 0.503 | 2026-07-09 | evidence |
| GPT-5.6 Sol | OpenAI | 0.527 | 2026-07-09 | evidence |
| GPT-5.6 Terra | OpenAI | 0.504 | 2026-07-09 | evidence |
| Qwen3.8 Max | Qwen | 0.524 | 2026-08-02 | evidence |
| Qwen3.8-27B | Qwen | 0.429 | 2026-08-14 | evidence |
| Seed 2.1 Pro | ByteDance | 0.414 | 2026-06-24 | 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.