SDABench - Inferential: leaderboard

Metric: Accuracy (%) of open-ended answers on the held-out SDA-Synth test split (the 400 inferential 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: Around December 2026. 13 models tracked.

Top models

#ModelScore
1GPT-5.467.25
2Gemini 3.1 Pro (Preview)65.5
3GLM-564.75
4Claude Sonnet 4.664.5
5Qwen 3.5 397B A17B61.5
6DeepSeek V3.259
7DeepSeek R146.5
8Kimi K2.543.5
9GPT-5.4 Mini40.25
10Qwen 3 235B A22B 2507 Instruct34.5
11Llama 3.3 70B Instruct8
12Llama 3.1 8B Instruct3.5

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