List your company

Financial numerical reasoning on DocMath Simplong (test)

70AccuracyMOCA-AGENT
50.2455.3760.565.63Jun 10, 2026

Evaluation results

MethodLinks
MOCA-AGENTBackbone=Qwen3.6-27B, Type=Open-source LLMs2026.0670
GPT-4oType=Proprietary LLMs2026.0660
GPT-o3-miniType=Proprietary LLMs2026.0659
GPT-4.5Type=Proprietary LLMs2026.0659
Qwen2.5-72BType=Open-source LLMs2026.0659
GPT-o1-previewType=Proprietary LLMs2026.0656
Qwen2.5-32BType=Open-source LLMs2026.0656
Fino1-7BType=Open-source LLMs2026.0656
Fino1-14BType=Open-source LLMs2026.0655
Llama-3.3-70BType=Open-source LLMs2026.0654
DeepSeek-V3Type=Proprietary LLMs2026.0653
DeepSeek-R1Type=Proprietary LLMs2026.0653
DS-R1-Distill-Llama-70BType=Open-source LLMs2026.0653
Qwen3-32BType=Open-source LLMs2026.0651