DPrivBench - Exponential Mechanism: leaderboard

Differential privacy verification for the report-noisy-max with Gumbel noise (the exponential mechanism) on the mechanism-level instances of DPrivBench (2026), where the model decides whether a stated differential privacy guarantee holds for a textbook mechanism instantiated with a concrete query, no tools, five seeds. Accuracy in percent; higher is better.

Metric: Accuracy (0-1, times 100) on the report-noisy-max with Gumbel noise (exponential mechanism) instances of the DPrivBench Category 1 mechanism-level instances (a mechanism template instantiated with a concrete query function from a 49-function bank; positive instances calibrate the noise to the true sensitivity, paired negatives under-calibrate it); binary yes/no verdict on whether the stated differential privacy guarantee holds, five seeds, no tools; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5High100#178
2Gemini 3 Pro95.9#145
3DeepSeek V3.1Non-reasoning81.8#511
4DeepSeek R176.3#482
5Gemini 2.5 Flash74.9#463
6Claude Sonnet 4.558.8#261
7Qwen 3 30B A3B 2507 Instruct52#1100
8Claude Opus 4.550.2#160
9Goedel-Prover-V2-32B50
10Qwen 3 30B A3B 2507 (Thinking)41.8#764

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=dprivbench-exponential-mechanism · How It Works · Data refreshed daily, snapshot 2026-11-21.