Benchmarks

Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.

309,131 indexed

Task	Dataset	Domain	Results	Last Update
Hallucination Detection	HaluEval, HotpotQA, TriviaQA, SQuAD	llm chat instructionnlp	36	Jun 30, 2026
Vulnerable code generation	Vulnerable code generation scenario	code developer aiadversarial safety security	36	Jun 30, 2026
Aerodynamic Coefficient Prediction	AIAA UQ Challenge 2 (withheld val points)	tabular time series	36	Jun 30, 2026
Long-context Reasoning	LongReason Average across 16k-64k	llm chat instruction	36	Jun 26, 2026
Long-context Reasoning	LongReason 64k prefill length