PatRe - Office Action Decision (OA-RS): leaderboard

PatRe (2026) models patent examination on 480 recent USPTO cases across all eight IPC sections. Acting as the examiner, the model writes an office action from the claims together with a noisy pool of retrieved references mixed with the cited ones (retrieval simulated). The score is the percent of office actions whose decision (allowance, non-final or final rejection, Ex parte Quayle) matches the examiner; higher is better.

Metric: Decision accuracy (%): the generated office action's decision (allowance, non-final or final rejection, Ex parte Quayle) matches the examiner's, office action generation under retrieval simulated: the model receives a noisy pool of BM25-retrieved references mixed with ground-truth ones, 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: Around September 2027. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Flash52.8#463
2GPT-5 Mini52.7#364
3Qwen 3.5 27B50.5#375
4Qwen 3.5 9B43.4#828
5DeepSeek V3.242.2#419
6Gemma 3 12B (IT)41.6#1887
7Llama 3.1 8B Instruct39.9#4600
8Gemma 3 27B (IT)38.1#1324
9GPT-4o Mini36.3#1667
10Llama 3.3 70B Instruct21.8#1385

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

Interactive version: theaggregate.ai/benchmark?slug=patre-office-action-decision-oa-rs · How It Works · Data refreshed daily, snapshot 2026-11-21.