SDABench - Descriptive: leaderboard

Metric: Accuracy (%) of open-ended answers on the held-out SDA-Synth test split (the 400 descriptive analysis instances; zero-shot, JSON reasoning and answer fields, the answer scored by relative-error numeric, exact or structured matching with a GPT-4o fallback judge); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 13 models tracked.

Top models

#ModelScore
1GPT-5.494.75
2Gemini 3.1 Pro (Preview)94.5
3Claude Sonnet 4.693
4Qwen 3.5 397B A17B89
5DeepSeek V3.288.5
6GLM-587.75
7DeepSeek R185
8Kimi K2.584.5
9GPT-5.4 Mini81.25
10Qwen 3 235B A22B 2507 Instruct77.25
11Llama 3.3 70B Instruct12
12Llama 3.1 8B Instruct10.25

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