JFinBench - No Tools - Financial Counter: leaderboard

321examples, 25sections, 6examfamilies; mean3runs. OneAPIcall/question, 131072outputtokens(65536forGemini), nativeeffort/thinking retained. MCQ extraction; GPT5.1 judging freeanswers undernativepartialcreditrubrics; exammean overexamples, overallmeanover6exams. Failure attempts0 pernativepolicy. Public321taskset reconstructablefromofficialsourcePDFrecipes; HFaloneomits57examples/55answers. Metric: Score (%). Source: https://arxiv.org/html/2611.18130v1

Metric: Score (%). Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked. Updated 17 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-6 AstraLow100#8
2Claude Fable 5Max100#28
3Gemini 3.8 FlashHigh98.5#39
4Kimi K3.1 ProMax90.9#32
5Kimi K3Low86.9#59
6DeepSeek V4.1 Flash (1017)Low82.3#44
7MiMo-V2.6-ProThinking76.3#86
8Mistral Large 4Thinking76.3#143
9Kimi K2.6Thinking74.8#191
10Qwen 3.8 FlashThinking73.7#128
11GLM-5.3 FlashMax65.7#136
12MiMo-V2.6-FlashThinking62.1#176

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=jfinbench-no-tools-financial-counter · How It Works · Data refreshed daily, snapshot 2026-11-17.