List your company

ResearchFormerly WizWand

Reasoning

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

GPQA Diamond (test)

34.85Accuracy

4

Apr 21, 2026

Overall Aggregate

90.1Accuracy

4

Apr 10, 2026

Multi-technique accuracy benchmark

3Math Accuracy

4

Mar 26, 2026

76.6Pass@1 Score

4

Mar 26, 2026

GRAFITE Sample Reasoning

Llama-3.3-70B-Instruct

83.3Pass Rate

4

Mar 20, 2026

37.09Inference Time (seconds)

4

Mar 9, 2026

60.93Inference Time (s)

4

Mar 9, 2026

31.7Inference Time (s)

4

Mar 9, 2026

Mathematical Reasoning (train)

1,177.8MTL (Loss)

4

Mar 4, 2026

GRPO-likelihood

0.26Distinct-1

4

Mar 4, 2026

ARC Challenge (test)

28Normalized Accuracy

4

Feb 26, 2026

ARC Easy (test)

53Normalized Accuracy

4

Feb 26, 2026

BigBench Extra Hard

14.3mean@4

4

Feb 26, 2026

Qwen3-1.7B-ALLMEM

67.3Accuracy

4

Feb 26, 2026

0.738Accuracy

4

Feb 26, 2026

76.76Accuracy

4

Feb 26, 2026

Composite (BIG-Bench Hard, BIG-Bench Extra Hard, bbh-zh, MuSR, ZebraLogic, PrOntoQA, PIQA, OCNLI, HellaSwag, KOR-Bench, DROP, SQuAD 2.0)

83.7BBH

4

Feb 26, 2026

Benchmark A BBH and paradoxes 1.0 (test)

94.7Accuracy

4

Feb 26, 2026

Qwen3-8B-thinking

80Accuracy

4

Feb 26, 2026

Shuffled Objects zero-shot

In-Writing-Base

50.8Accuracy

4

Feb 26, 2026

Last Letter Concatenation zero-shot

70.1Accuracy (Zero-shot)

4

Feb 26, 2026

GSM8K zero-shot

In-Writing-Base

86.9Accuracy

4

Feb 26, 2026

Reasoning Hop Generalization Tasks Parity-NL, MDM, LLC, CLF, MOAS, ObjC, NumS (test)

87.7Parity-NL

4

Feb 26, 2026

MMLU STEM (out-of-distribution)

Qwen3-4B-Original

93.3Accuracy

4

Feb 26, 2026

75.3MMLU-Pro Accuracy

4

Feb 26, 2026