OpenAI's GPT-6.1 Sol nears Astra at quarter the cost
OpenAI's GPT-6.1 Sol replaced GPT-6 Sol after just seven days, and Artificial Analysis measures it one point below GPT-6 Astra on its Intelligence Index at less than a quarter of the cost per task.
OpenAI has replaced GPT-6 Sol with GPT-6.1 Sol only seven days after the earlier model shipped. Independent benchmarking firm Artificial Analysis reports that the new model scores 4 points higher than GPT-6 Sol on its Intelligence Index, putting it 1 point below GPT-6 Astra.
At max effort, Artificial Analysis puts GPT-6.1 Sol's cost per Intelligence Index task at $0.72. That compares with $3.26 for GPT-6 Astra, $1.05 for GPT-6 Sol and $1.99 for GPT-5.6 Sol. The firm says every effort level of the model sits on the cost-efficiency Pareto frontier, meaning no cheaper model matches it at a given intelligence level.
List pricing stays at $2/$10 per million input/output tokens. The cache-read discount rises from 90% to 95%, which slightly lowers blended prices for agentic workloads. GPT-6 Sol had already cut prices by 50% compared with GPT-5.6 Sol.
The biggest gains reported by Artificial Analysis:
- Terminal-Bench 4.0: up 12 points
- AA-Omniscience Accuracy: up 8 points, with the hallucination rate falling from 60% to 54%
- GDP.pdf: up 6 points
- Humanity's Last Exam: up 5 points
- GDPval-AA v2.1: up 5 points
- AA-Briefcase v1.1: up 4 points, roughly 80 Elo
On the Coding Agent Index, the xhigh setting scores 1 point above GPT-6 Astra for under 15% of its cost per task. The xhigh setting beat max effort by 3 points.
There is one tradeoff: GPT-6.1 Sol uses about 10–30% more output tokens than GPT-6 Sol. Even so, its low and medium effort settings are still Pareto-optimal for token efficiency.
OpenAI has not said why it replaced the model so quickly.