Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 1172 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Line Retrieval | Line Retrieval 100 lines | 24 | Jun 4, 2026 | ||
| Instruction Following | IFEval | 24 | Jun 4, 2026 | ||
| LLM Alignment | Ultrafeedback | 24 | Jun 11, 2026 | ||
| Preference Alignment | Skywork | 24 | Jun 4, 2026 | ||
| Causal Inference | Kang & Schafer | 24 | Jun 4, 2026 | ||
| Mathematical Reasoning | AIME24 | 24 | Jun 30, 2026 | ||
| Opinion Editing | FOE 1.0 (test) | 24 | Jun 4, 2026 | ||
| Multi-component causal tracing | CounterFact Factual | 24 | Jun 4, 2026 | ||
| Attention Head Selection (Causal Tracing) | WinoBias | 24 | Jun 4, 2026 | ||
| Attention Head Selection (Causal Tracing) | WinoGender | 24 | Jun 4, 2026 | ||
| Inertial Odometry | Aria (test) | 24 | Jun 4, 2026 | ||
| Inertial Odometry | Nymeria (test) | 24 | Jun 4, 2026 | ||
| Deepfake Detection | Adversarial Fake Images SP24 | 24 | Jun 4, 2026 | ||
| Memory recall | noisy_long_recall primary memory benchmark | 24 | Jun 4, 2026 | ||
| Multi-Hop Question Answering | Bamboogle (test) | 24 | Jul 7, 2026 | ||
| Multi-Hop Question Answering | 2Wiki (test) | 24 | Jul 7, 2026 | ||
| Price Publication | Price publication problem Linear regime, q=1, 1000 settlement periods | 24 | Jun 2, 2026 | ||
| Agentic Safety and Utility Evaluation | PowerSeeking Bench | 24 | Jun 2, 2026 | ||
| Continual Instruction Tuning | TRACE | 24 | Jun 2, 2026 | ||
| Continual Instruction Tuning | NI-Seq M1 | 24 | Jun 2, 2026 | ||
| Continual Instruction Tuning | NI-Seq-C1 | 24 | Jun 2, 2026 | ||
| Continual Instruction Tuning | NI-Seq-G1 | 24 | Jun 2, 2026 | ||
| Part-level Pose Estimation | Real-world dataset | 24 | Jun 2, 2026 | ||
| Articulated Joint Estimation | Real-world dataset | 24 | Jun 2, 2026 | ||
| Part Segmentation | Real-world dataset | 24 | Jun 2, 2026 |