Artificial Analysis

Artificial Analysis provides independent evaluation and comparison of large language models (LLMs) across multiple dimensions including intelligence benchmarks, speed metrics, cost efficiency, and quality assessments. The platform offers comprehensive benchmarking data covering over 300 AI models from major providers, including proprietary and open-source options.

The platform features the Artificial Analysis Intelligence Index (v3.0), which combines 10 evaluation metrics: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME 2025, IFBench, AA-LCR, Terminal-Bench Hard, and τ²-Bench Telecom. Additional specialized benchmarks include the AA-Omniscience Index for knowledge reliability and hallucination measurement, along with comprehensive speed, latency, and pricing comparisons across API providers.

All evaluations are conducted independently on dedicated hardware using standardized methodologies. The platform tracks model performance across intelligence, output speed, input/output pricing, cost efficiency, and API provider performance. Interactive visualizations enable direct comparison of frontier models, open-weight versus proprietary models, and reasoning versus non-reasoning architectures.