Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 2059 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Code Generation | MBPP | 17 | Feb 26, 2026 | ||
| Science Reasoning | GPQA Diamond | 17 | Feb 26, 2026 | ||
| Mathematical Reasoning | MATH500 | 17 | Feb 26, 2026 | ||
| Commonsense Reasoning | Commonsense Reasoning Benchmarks zero-shot LLaMA-2-13B | 17 | Feb 26, 2026 | ||
| Language Modeling | Perplexity Evaluation zero-shot | 17 | Feb 26, 2026 | ||
| Scientific Reasoning | OlympiadBench | 17 | Jul 3, 2026 | ||
| Question Answering | Average (PopQA, NQ, HotpotQA, 2Wiki, MuSiQue) | 17 | Feb 26, 2026 | ||
| Audio-to-Video temporal grounding | ChronusAV | 17 | Jul 7, 2026 | ||
| Video-to-Audio temporal grounding | ChronusAV | 17 | Jul 7, 2026 | ||
| Text-to-Audio temporal grounding | ChronusAV | 17 | Jul 7, 2026 | ||
| Long-context understanding | LongBench | 17 | Feb 26, 2026 | ||
| Mathematical Reasoning | CARP-EN | 17 | Jun 30, 2026 | ||
| Mathematical Reasoning | AIME 2025 | 17 | Feb 26, 2026 | ||
| Multiple Choice Question Answering | Bolmo Evaluation Suite MC STEM 7B | 17 | Feb 26, 2026 | ||
| Metric Correlation Analysis | Synthetic perturbation sets (test) | 17 | Feb 26, 2026 | ||
| Machine Translation | WMT En–Ja 2023 (dev) | 17 | Feb 26, 2026 | ||
| Jailbreak Attack | HarmBench example-based Llama3 8B | 17 | Apr 16, 2026 | ||
| Summarization | QMSum (val) | 17 | Feb 26, 2026 | ||
| Summarization | GovReport (val) | 17 | Feb 26, 2026 | ||
| Agentic Instruction Following | TaskCraft | 17 | Feb 26, 2026 | ||
| Safety Evaluation | X (test) | 17 | Jun 23, 2026 | ||
| Proactive next utterance prediction | CrossWOZ (test) | 17 | Feb 26, 2026 | ||
| Proactive next utterance prediction | WildChat (test) | 17 | Feb 26, 2026 | ||
| Proactive next utterance prediction | ShareGPT (test) | 17 | Feb 26, 2026 | ||
| Proactive next utterance prediction | LMSYS (test) | 17 | Feb 26, 2026 |