Can AI replace financial analysts? Vals AI's new version fails miserably in tests, with GPT 5.5 barely achieving over 50% accuracy.

According to Beating, an AI testing organization, Vals AI released its second-generation Financial Agent benchmark test (Finance Agent v2). This is an end-to-end test simulating the workflow of a junior financial analyst, containing 927 expert-reviewed questions. The new version of the test is significantly more difficult, with GPT 5.5 only achieving a 51.76% accuracy rate, extremely close to Claude Opus 4.7 (51.51%) and Claude Sonnet 4.6 (51.03%). Unlike single-round question-and-answer tests, this test requires the model to autonomously find relevant passages in hundreds of pages of 10-K and 10-Q financial statements, handle cross-year financial statement adjustments, and complete multi-step calculations with precise intermediate figures. Vals AI revealed that if a strict scoring standard of "must answer completely correctly" is adopted, the accuracy of all cutting-edge models drops below 40%; in the most difficult categories of "financial modeling" and "precedent analysis," the highest score is only 23%. In terms of other models, Kimi K2.6 ranked fifth with 44.87%, making it the highest-scoring domestic model; followed by GLM 5.1 (44.79%) and DeepSeek V4 (44.08%). Furthermore, the official "fastest speed" label was awarded to Claude Opus 4.7 (single time 360 seconds), while GLM 5.1 received the "most cost-effective" label (single cost $0.62). The collective drop in scores in this test (Opus 4.7 scored 64.4% in the previous generation test) proves one point: current AI can handle simple retrievals, but in the complex areas of finance where specific industry practices and extremely high numerical accuracy are required, it is far from replacing human analysts.