As of August 17, 2026, reports indicate a 50% pricing reduction for GPT-5.6 Sol. Technical benchmarking data from Artificial Analysis reveals significant performance variance across different model configurations, ranging from "low" to "max" settings. High-tier configurations demonstrate superior capabilities, with the "max" variant achieving an Intelligence Index of 47.0 and a Coding Index of 77.4. In advanced reasoning tasks, the "max" model reaches a GPQA Diamond score of 94.1% and an HLE score of 49.5%, while the "low" version performs notably lower with a 33.5 Intelligence Index and an 89.8% GPQA Diamond score. Agentic capabilities also scale with model intensity, moving from a 31.6 index in "low" to 50.2 in "max" configurations. Conversely, the "Non-reasoning" version shows diminished performance across all metrics, including a 65.1 Coding Index and a significantly lower HLE score of 16.7%. Other notable metrics for the "max" tier include an IFBench score of 72.7% and an AA-LCR of 84.0%, whereas the "low" tier maintains an IFBench score of 66.5% and an AA-LCR of 78.0%.