GPT-6.1 Sol Replaces GPT-6 Sol with Near-Astra Intelligence at Lower Cost

OpenAI has replaced GPT-6 Sol with GPT-6.1 Sol just seven days after the previous version's release. The new model achieves near-Astra intelligence levels, scoring only 1 point below GPT-6 Astra in the Artificial Analysis Intelligence Index, while maintaining the same base pricing as GPT-6 Sol.

Cost and Efficiency

GPT-6.1 Sol pushes the cost efficiency frontier significantly. At max effort, it costs $0.72 per Intelligence Index task, which is less than a quarter of GPT-6 Astra's $3.26 and 31% less than GPT-6 Sol's $1.05. While the base price remains $2/$10 per million input/output tokens, the cache read discount increases from 90% to 95%, lowering the blended price for agentic workloads.

Performance Gains

The model shows substantial improvements across key benchmarks compared to its predecessor:

• Intelligence Index: +4 points vs GPT-6 Sol, +5 points vs GPT-5.6 Sol.
• Agentic Knowledge Work: +4 points in AA-Briefcase v1.1, +5 points in GDPval-AA v2.1.
• Coding: +3 points in the Artificial Analysis Coding Agent Index at max effort; the xhigh effort setting outperforms max by 3 points.
• Accuracy: +8 points in AA-Omniscience Accuracy, with hallucination rates dropping from 60% to 54%.
• Other Metrics: +12 points in Terminal-Bench 4.0, +5 points in Humanity’s Last Exam, and +6 points in GDP.pdf.

Token Usage

GPT-6.1 Sol uses 10-30% more output tokens than GPT-6 Sol across effort levels. However, due to the higher intelligence scores, its low and medium effort settings remain Pareto optimal for token efficiency.