AA-AnalystAgent

A data analysis benchmark, testing AI agents on their ability to work with spreadsheets and documents to answer quantitative questions a Business Analyst or Data Analyst would face day-to-day.

Models: 26 current, 52 in total
Median: 40.0%
Range: 1.3%–57.5%

Score distribution

Lowest: 1.3%
Median: 40.0%
Highest: 57.5%

Leaderboard

Include deprecated models

Rank	Model	Score
1	Claude Sonnet 5.5 (Max, Default Fallback) Anthropic	57.5%
2	Claude Opus 5.5 (Max, Default Fallback) Anthropic	56.3%
3	Grok 4.7 (High) SpaceXAI	55.0%
4	Claude Haiku 5.5 (Max) Anthropic	53.8%
5	Grok 4.7 (Xhigh) SpaceXAI	52.5%
6	GPT-6 Astra (Max) OpenAI	51.2%
7	Inkling 2 Small Thinking Machines	51.2%
8	GPT-6.1 Sol (Max) OpenAI	50.0%
9	Gemini 4 Argon (High) Google	48.8%
10	Gemini 3.8 Flash (High) Google	46.3%
11	Qwen3.8 Max (0902) Alibaba	45.0%
12	GPT-6 Luna (Max) OpenAI	43.8%
13	Gemini 3.1 Pro Preview Google	41.3%
14	GLM-5.3 (Max) Z AI	38.8%
15	Kimi K3.1 Pro Kimi	38.8%
16	MiMo-V2.6-Pro Xiaomi	37.5%
17	Step 5 Preview StepFun	35.0%
18	Inkling Small Thinking Machines	27.5%
19	GLM-5.3-Flash Z AI	25.0%
20	Inkling (Xhigh) Thinking Machines	23.8%
21	Muse Glimmer (High) Meta	20.0%
22	Ling-3.0-flash-Fin InclusionAI	16.3%
23	Mistral Medium 3.5 Mistral	12.5%
24	MiniMax-M3 MiniMax	10.0%
25	Nemotron 3 Ultra 550B A55B (Reasoning) NVIDIA	6.3%
26	Mistral Small 4 (Reasoning) Mistral	1.3%

1
Claude Sonnet 5.5 (Max, Default Fallback)
Anthropic
57.5%
2
Claude Opus 5.5 (Max, Default Fallback)

Data from Artificial Analysis

Command Palette

AA-AnalystAgent

Score distribution

Leaderboard