Opus 5.5 still leads on AA intelligence, but GPT-6.1 Sol shifts the cost frontier
retained text
This compares reasoning effort settings using Artificial Analysis's Intelligence Index and cost per benchmark task. Costs include caching and reasoning; they aren't subscription prices. Data source: https://artificialanalysis.ai/
document fetched 2026-09-30 00:29 · canonical https://reddit.com/r/OpenAI/comments/1wtksk2/opus_55_still_leads_on_aa_intelligence_but_gpt61