Benchmark
APEX-Agents
Expert-authored professional tasks with rubric grading; graders are LLMs.
- Released
- 2026-01-20
- Categories
- agent
- Openness
- unknown
- Source
- model_reports
- Reported scores
- 5
Reported scores
model_reports
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| DeepSeek-V4-Flash | DeepSeek | 33.0 | 2026-06-25 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 38.3 | 2026-04-22 | evidence |
| Gemini 3.1 Pro | 33.5 | 2026-02-19 | evidence | |
| Hy4 preview | Tencent | 37.1 | 2026-08-28 | evidence |
| Kimi K3 | Moonshot AI | 41.0 | 2026-06-13 | 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.