SDABench - Exploratory (MCQ): leaderboard

Metric: Accuracy (%) of four-option multiple-choice answers with perturbation-based distractors on the held-out SDA-Synth test split (the 400 exploratory analysis instances; zero-shot; 25% chance); higher is better. Source: arxiv.org. Saturation forecast: Around June 2028. 13 models tracked.

Top models

#ModelScore
1Claude Sonnet 4.651.75
2Gemini 3.1 Pro (Preview)49
3GPT-5.445.5
4DeepSeek V3.244
5Qwen 3.5 397B A17B44
6DeepSeek R142.25
7GLM-538.75
8Qwen 3 235B A22B 2507 Instruct37.75
9GPT-5.4 Mini36.25
10Kimi K2.534.5
11Llama 3.3 70B Instruct23.5
12Llama 3.1 8B Instruct21.5

Interactive version: theaggregate.ai/benchmark?slug=sdabench-exploratory-mcq · How It Works · Data refreshed daily, snapshot 2026-09-29.