PatRe - Rebuttal Coverage: leaderboard

PatRe (2026) models patent examination on 480 recent USPTO cases across all eight IPC sections. Acting as the applicant, the model writes the rebuttal to an office action; the score is the percent of the atomic rejection points of the office action that the rebuttal answers, higher is better.

Metric: Point-wise coverage (%): share of the atomic rejection points of the office action that the generated applicant rebuttal answers, over PatRe (480 recent USPTO patent examination histories across all eight IPC sections, with office actions, rebuttals, claim versions and cited references), temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5 Mini90.5#364
2Gemini 2.5 Flash81.5#463
3Qwen 3.5 27B79.2#375
4DeepSeek V3.279.1#419
5Qwen 3.5 9B63.4#828
6Gemma 3 27B (IT)39.9#1324
7Llama 3.3 70B Instruct30.9#1385
8Gemma 3 12B (IT)30.5#1887
9Llama 3.1 8B Instruct29.5#4600
10GPT-4o Mini27.8#1667

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

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