SPADE Zero Shot - RAID Poetry AUROC: leaderboard

The SPADE paper compares six zero-shot AI-text detectors using three released language-model backbones. AUROC and true-positive rate at 5% false positives remain separate. Paired base/instruct checkpoints are retained for Binoculars and Fast-DetectGPT. Learned soft prompts and supervised detector weights are distinct artifacts. This board reports aUROC (%). Higher is better.

Metric: AUROC (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 18 models tracked. Updated 14 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1SPADE Binoculars: Llama 3.1 8B + Llama 3.1 8B InstructEvaluation run: frozen two-checkpoint detector91.6
2SPADE Binoculars: Falcon-7B + Falcon-7B-InstructEvaluation run: frozen two-checkpoint detector90.3
3falcon-7B (Harness: SPADE zero-shot NTS)89.9
4SPADE Fast-DetectGPT: Llama 3.1 8B + Llama 3.1 8B InstructEvaluation run: frozen two-checkpoint detector87.5
5SPADE Fast-DetectGPT: Falcon-7B + Falcon-7B-InstructEvaluation run: frozen two-checkpoint detector84.9
6falcon-7B (Harness: SPADE zero-shot DNA-GPT)77
7Qwen3-8B-Base (Harness: SPADE zero-shot Likelihood)75.3
8Qwen3-8B-Base (Harness: SPADE zero-shot Entropy)75.3
9falcon-7B (Harness: SPADE zero-shot Likelihood)72.1
10Llama 3.1 8B (Harness: SPADE Zero Shot Likelihood)70.3
11Llama 3.1 8B (Harness: SPADE Zero Shot Entropy)68.9
12Qwen3-8B-Base (Harness: SPADE zero-shot DNA-GPT)68.3
13Llama 3.1 8B (Harness: SPADE Zero Shot NTS)67.9
14Llama 3.1 8B (Harness: SPADE Zero Shot DNA GPT)67.1
15SPADE Binoculars: Qwen3-8B-Base + Qwen3-8BEvaluation run: frozen two-checkpoint detector64.9

Interactive version: theaggregate.ai/benchmark?slug=spade-zero-shot-raid-poetry-auroc · How It Works · Data refreshed daily, snapshot 2026-11-16.