Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 2065 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Concept Unlearning | I2P | 17 | Feb 26, 2026 | ||
| Reasoning | HMMT | 17 | Feb 26, 2026 | ||
| Scientific Figure Reasoning | CharXiv | 17 | Feb 26, 2026 | ||
| General Visual Reasoning | M3CoT | 17 | Feb 26, 2026 | ||
| Agentic Reasoning | GAIA (val) | 17 | Feb 26, 2026 | ||
| Terminal task completion | Terminal-bench 1.0 | 17 | Feb 26, 2026 | ||
| Proactive Agent Task | Turtle-Gym | 17 | Feb 26, 2026 | ||
| Proactive Agent Task | Telepathy-Gym | 17 | Feb 26, 2026 | ||
| Proactive Agent Task | Function-Gym | 17 | Feb 26, 2026 | ||
| Visual Grounding | V* Bench | 17 | May 18, 2026 | ||
| Facial Expression Recognition | RML (test) | 17 | Feb 26, 2026 | ||
| Sentiment Analysis | SST2 | 17 | Feb 26, 2026 | ||
| Spoken Dialogue State Tracking | MultiWOZ (test) | 17 | Feb 26, 2026 | ||
| Diabetic Retinopathy (DR) grading | RLDR | 17 | Feb 26, 2026 | ||
| Text-to-Video Retrieval | MSRVTT | 17 | Feb 26, 2026 | ||
| Web Search Retrieval | MS MARCO (dev) | 17 | Feb 26, 2026 | ||
| Dynamic Scene Reconstruction | Neural 3D Video 19 (full) | 17 | Feb 26, 2026 | ||
| Single-hop Question Answering | Factrecall en | 17 | Feb 26, 2026 | ||
| Single-hop Question Answering | Loogle SD | 17 | Feb 26, 2026 | ||
| Correlation with human evaluation | SEEDA Fluency (E) | 17 | Feb 26, 2026 | ||
| Correlation with human evaluation | SEEDA Fluency (S) | 17 | Feb 26, 2026 | ||
| Correlation with human evaluation | SEEDA-S Base | 17 | Feb 26, 2026 | ||
| Event Argument Extraction | CASIE | 17 | Apr 14, 2026 | ||
| Harmlessness | Toxigen | 17 | Feb 26, 2026 | ||
| Image Classification | Classical SSL Benchmarks Overall (aggregated) | 17 | Feb 26, 2026 |