FrontierFinance Benchmark Tests AI Models on Financial Intelligence

OpenAI's GPT 5.6 Sol was evaluated on FrontierFinance, a newly released benchmark for financial intelligence launched alongside the model. The model achieved 46.8% accuracy, trailing Fable 5 at 49.2% and outpacing Opus 4.8 at 45%. While ranking third in quality, GPT 5.6 Sol demonstrated cost advantages per query compared to competitors. Samaya's Light system occupies the Pareto frontier with the highest quality score (50.8%) at the lowest cost, approximately four times cheaper than alternatives. The evaluation revealed qualitative differences between models in analytical capabilities and financial data extraction. FrontierFinance positions itself as the most comprehensive and rigorous public benchmark for assessing financial AI intelligence.