Edison Advances
Benchmarks / LABBench2

PatentQA

patentqa

10 runs · 5 models · evaluated by Score.

# Model ↕ Variant ↕ Mode ↕ Average score ↓ Average output tokens ↕ Avg. dur ↕ Date ↕
1 gpt-5-2-pro tools,high inject 0.909 3.3k 4.3m 2026-02-03
2 gpt-5-2 tools,high inject 0.802 3.5k 2.0m 2026-02-03
3 claude-opus-4-6 tools,high inject 0.570 2.5k 3.6m 2026-03-22
4 claude-opus-4-5 tools,high inject 0.438 816 2.4m 2026-03-22
5 gemini-3-pro-preview tools,high inject 0.438 446 3.2m 2026-02-03
6 gemini-3-pro-preview — inject 0.215 313 1.1m 2026-02-03
7 gpt-5-2-pro — inject 0.182 974 1.2m 2026-02-03
8 claude-opus-4-6 — inject 0.174 273 8.5s 2026-03-20
9 gpt-5-2 — inject 0.165 120 3.8s 2026-02-03
10 claude-opus-4-5 — inject 0.099 236 6.7s 2026-03-20

Click column headers to sort. Click mode chips to filter.