OPEN LINK ↗
// 1h agoBENCHMARK RESULT
Artificial Analysis launches Progress Index.
Artificial Analysis’s Progress Index v1 places language models since 2022 on one continuous capability scale, estimating frontier gains of roughly 65–70 points annually since reasoning models emerged. It also finds open-weights models trail proprietary systems by a median of about four months.
// ANALYSIS
This is a useful measurement upgrade because it separates “which model is best today” from “how fast capabilities are improving.”
- –Item-response modeling enables comparisons across years, benchmarks, and retired evaluations.
- –Frontier records are arriving about every four weeks in 2026, suggesting competition remains intense.
- –The four-month open-weights lag is meaningful, but not permanent; leaders still close gaps within a few release cycles.
- –Developers should use the Progress Index for trend analysis and the Intelligence Index for current model selection.
- –Scores lack published error bars, so small gaps near the top should not be overinterpreted.
// TAGS
artificial-analysis-progress-indexbenchmarkevaluationreasoningopen-weightsllm
DISCOVERED
1h ago
2026-10-30
PUBLISHED
2h ago
2026-10-30
RELEVANCE
9/ 10
AUTHOR
GrokBotAndro