GoKawiilArtificial Analysis releases Intelligence Index v5 with new benchmarks and private datasets
Artificial Analysis launched version 5 of its Intelligence Index, a composite benchmark measuring AI model capabilities, adding two new tests (FrontierCode 2.0 and Terminal-Bench-Science) and upgrading three others, including a refined version of Humanity's Last Exam. The firm also introduced a companion Progress Index to track model capability over time, and reports that 60% of the Index now relies on private or held-out test data. In current results, Claude Opus 5.5 leads with a score of 58, narrowly ahead of GPT-6 Astra at 57.
GoKawiil's interpretation of the reporting above, not reported fact.
Shifting most of the benchmark to private datasets suggests an effort by Artificial Analysis to reduce the risk of AI labs training models specifically to game public test sets, which has been a recurring criticism of benchmark-based rankings. The addition of agentic, tool-using and science-focused tasks indicates the industry's evaluation standards are moving beyond simple question-answering toward tasks resembling real workplace and research activity. Because these indices increasingly shape public perception of which models are 'best,' changes to their methodology can meaningfully affect competitive narratives among labs like OpenAI and Anthropic.
- Intelligence Index v5 adds FrontierCode 2.0 and Terminal-Bench-Science while upgrading HLE-Diamond, AA-Omniscience, and CritPt v2.
- 60% of the Index is now based on private or held-out datasets, up significantly from prior versions.
- Claude Opus 5.5 currently tops the index at 58, just ahead of GPT-6 Astra at 57.
Source: artificialanalysis.ai, 2026-10-28
Published there as: “Artificial Analysis Intelligence Index v5”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.