List your company

ResearchFormerly WizWand

Reasoning evaluation

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

99.1Reasoning

33

Feb 26, 2026

DeepfakeJudge Reason 1.0 (test)

Qwen-3-VL-30B-Instruct

9BLEU-1

16

Feb 26, 2026

Qwen2.5-Math-PRM-7B-PDDL-r

56.1Math F1

10

Apr 21, 2026

63.7WR vs. Base

8

May 11, 2026

109-sample (test)

Universe Routing

97.25Accuracy

7

Mar 17, 2026

Limit Texas Hold’em

2Hit Rate

6

Feb 26, 2026

Leduc Hold’em

2Hit Rate (HR)

6

Feb 26, 2026

Full task set (n=45)

8.84Overall Score

5

Mar 13, 2026