Kaggle Can AI Close the Books (FP&A Reasoning): leaderboard

Kaggle Benchmarks corporate finance-close evaluation (marcopolofpa/can-ai-close-the-books-fp-and-a-reasoning): 28 deterministic synthetic FP&A reasoning tasks covering penny-reconciled variance bridges, accruals and cut-off adjustments, three-statement linkage, and valid/invalid reviewer claims. Exact pass rate (0-100%); higher is better.

Metric: Pass Rate (%). Source: www.kaggle.com. Saturation forecast: Around February 2027. 40 models tracked. Updated 21 Nov 2026.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Pro100#276
2Gemini 3.1 Pro (Preview)100#99
3GPT-5.5100#54
4GPT-5.6 Sol100#37
5Gemini 3 Flash (Preview)100#152
6Gemma 4 31B100#382
7GLM-5100#269
8Claude Opus 5100#26
9Gemini 3.8 Flash100#43
10Claude Opus 5.5100#9
11GPT-5.6 Luna100#131
12Gemma 4 26B A4B100#590
13Claude Sonnet 5100#118
14GPT-6 Luna100#105
15Gemini 3.6 Flash100#81

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=kaggle-can-ai-close-the-books-fp-a-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-21.