AIDABench
Agents analyze documents and spreadsheets to answer questions, draw charts, and generate files across 603 tasks. They can execute Python in a sandbox. Separate model evaluators check answer correctness, chart correctness and readability, and spreadsheet structure and content.
About 27% of tasks require multiple files. The benchmark uses different graders for its three output types. The paper was submitted in February 2026 despite its March arXiv identifier.
- Made by
- SenseTime Research and collaborators
- Released
- February 2026
- Kind
- Documents and spreadsheets
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- Code sandbox
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Agents, Image, video and 3D generation
- Strongest model tested
-
Gemini 3 Pro #56 on the capability index
- Last model added
- January 2026
- Scores refreshed
- November 20th, 2026
In the authors’ words
As AI-driven document understanding and processing tools become increasingly prevalent in real-world applications, the need for rigorous evaluation standards has grown increasingly urgent. Existing benchmarks and evaluations often focus on isolated capabilities or simplified scenarios, failing to capture the end-to-end task effectiveness required in practical settings. To address this gap, we introduce AIDABench, a comprehensive benchmark for evaluating AI systems on complex data analytics tasks in an end-to-end manner. AIDABench encompasses 600+ diverse document analysis tasks across three core capability dimensions: question answering, data visualization, and file generation. These tasks are grounded in realistic scenarios involving heterogeneous data types, including spreadsheets, databases, financial reports, and operational records, and reflect analytical demands across diverse industries and job functions. Notably, the tasks in AIDABench are sufficiently challenging that even human experts require 1-2 hours per question when assisted by AI tools, underscoring the benchmark's difficulty and real-world complexity. We evaluate 11 state-of-the-art models on AIDABench, spanning both proprietary (e.g., Claude Sonnet 4.5, Gemini 3 Pro Preview) and open-source (e.g., Qwen3-Max-2026-01-23-Thinking) families. Our results reveal that complex, real-world data analytics tasks remain a significant challenge for current AI systems, with the best-performing model achieving only 59.43% pass-at-1. We provide a detailed analysis of failure modes across each capability dimension and identify key challenges for future research. AIDABench offers a principled reference for enterprise procurement, tool selection, and model optimization, and is publicly available at https://github.com/MichaelYang-lyx/AIDABench.
Claude Sonnet 4.5 leads with 59.4% among models released by January 2026 Claude Sonnet 4.5 leads with 59.4% among models released by January 2026
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in January 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 Claude Sonnet 4.5 59.4%
- 2
Gemini 3 Pro 57.4%
- 3
Qwen3-Max-2026-01-23 56.7%
- 4 Kimi K2 Thinking 54.2%
- 5
DeepSeek-V3.2 53.8%
- 6
Qwen3-235B-A22B-Instruct Jul 2025 47.8%
- 7
Qwen3-30B-A3B-Thinking Jul 2025 38.7%
- 8 kimi-k2-turbo-preview 38.3%
- 9
Qwen3-30B-A3B-Instruct Jul 2025 31.4%
Source: The AIDABench paper
- 1 Claude Sonnet 4.5 59.4%
- 2
Gemini 3 Pro 57.4%
- 3
Qwen3-Max-2026-01-23 56.7%
- 4 Kimi K2 Thinking 54.2%
- 5
DeepSeek-V3.2 53.8%
- 6
Qwen3-235B-A22B-Instruct Jul 2025 47.8%
- 7
Qwen3-30B-A3B-Thinking Jul 2025 38.7%
- 8
Qwen3-30B-A3B-Instruct Jul 2025 31.4%
Source: The AIDABench paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 59.4% | Sep 2025 | Official leaderboard | |
| 2 | | 57.4% | Nov 2025 | Official leaderboard | |
| 3 | | 56.7% | Jan 2026 | Official leaderboard | |
| 4 | Kimi K2 Thinking | 54.2% | Nov 2025 | Official leaderboard | |
| 5 | | 53.8% | Dec 2025 | Official leaderboard | |
| 6 | | 47.8% | Jul 2025 | Official leaderboard | |
| 7 | | 38.7% | Jul 2025 | Official leaderboard | |
| 8 | kimi-k2-turbo-preview 1000B | 38.3% | Official leaderboard | ||
| 9 | | 31.4% | Jul 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 59.4% | Sep 2025 | Official leaderboard | |
| 2 | | 57.4% | Nov 2025 | Official leaderboard | |
| 3 | | 56.7% | Jan 2026 | Official leaderboard | |
| 4 | Kimi K2 Thinking | 54.2% | Nov 2025 | Official leaderboard | |
| 5 | | 53.8% | Dec 2025 | Official leaderboard | |
| 6 | | 47.8% | Jul 2025 | Official leaderboard | |
| 7 | | 38.7% | Jul 2025 | Official leaderboard | |
| 8 | | 31.4% | Jul 2025 | Official leaderboard |
Rankings by part
We read 3 parts from the benchmark’s paper: QA, Visualization and File generation. Each ranks models on one part of the test. Claude Sonnet 4.5 leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
QA
8 models we track have a score here.
- 1 Claude Sonnet 4.5 68.6%
- 2
Qwen3-Max-2026-01-23 ↑1 66.8%
- 3 Kimi K2 Thinking ↑1 66.4%
- 4
Gemini 3 Pro ↓2 64.6%
- 5
DeepSeek-V3.2 62.4%
- 6
Qwen3-235B-A22B-Instruct Jul 2025 54%
- 7
Qwen3-30B-A3B-Thinking Jul 2025 52.6%
- 8
Qwen3-30B-A3B-Instruct Jul 2025 35.4%
Visualization
8 models we track have a score here.
- 1 Claude Sonnet 4.5 67.7%
- 2
Qwen3-Max-2026-01-23 ↑1 64.4%
- 3
Gemini 3 Pro ↓1 63.9%
- 4
Qwen3-235B-A22B-Instruct Jul 2025 ↑2 63.1%
- 5
DeepSeek-V3.2 62.5%
- 6 Kimi K2 Thinking ↓2 60%
- 7
Qwen3-30B-A3B-Thinking Jul 2025 50.9%
- 8
Qwen3-30B-A3B-Instruct Jul 2025 47.6%
File generation
8 models we track have a score here.
- 1 Claude Sonnet 4.5 49.4%
- 2
Gemini 3 Pro 48.3%
- 3
Qwen3-Max-2026-01-23 44.8%
- 4
DeepSeek-V3.2 ↑1 43.3%
- 5 Kimi K2 Thinking ↓1 41%
- 6
Qwen3-235B-A22B-Instruct Jul 2025 36%
- 7
Qwen3-30B-A3B-Thinking Jul 2025 22.2%
- 8
Qwen3-30B-A3B-Instruct Jul 2025 21.5%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The AIDABench paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official AIDABench leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Economic work
- JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
- GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate.
Seed2.1 Pro 87.9%
- GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
- AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
- Agents' Last Exam v1.1 In use Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. GPT-6.2 Astra 47.5%
- CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 10 months ago, in January 2026, for Qwen3-Max-2026-01-23. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (9 entries).
- Pass Models scored 8 models we track, plus 1 other entry on its leaderboard.
- Warning Newest score The newest came 10 months ago, in January 2026, for Qwen3-Max-2026-01-23.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3 Pro (#56).
- Pass Headroom The best is 59.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 10 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.