Winogrande Benchmark Leaderboard | Phaseo

Winogrande

AI benchmark results and model performance

Type: percentageLanguage

Recorded Results

17

Average Score

67.68%

Score Range

51.30% - 87.50%

Leading Model (lowest score)

51.30% - Ernie 4.5 21B A3B

Scores Over Time

Individual benchmark scores plotted by date.

Models Using This Benchmark

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
Baidu	Ernie 4.5 21B A3B	-	51.30%	inferred version-family alias from ernie-4.5	Yes	Source
Baidu	Ernie 4.5 21B A3B Thinking	-	51.30%	inferred version-family alias from ernie-4.5	Yes	Source
Baidu	Ernie 4.5 300B A47B	-	51.30%	inferred version-family alias from ernie-4.5	Yes	Source
Baidu	Ernie 4.5 Turbo	-	51.30%	inferred version-family alias from ernie-4.5	Yes	Source
Baidu	Ernie 4.5 VL 28B A3B	-	51.30%	inferred version-family alias from ernie-4.5	Yes	Source
Baidu	Ernie 4.5 VL 424B A47B	-	51.30%	inferred version-family alias from ernie-4.5	Yes	Source
Google	Gemma 3n E2B	25 Jun 2025	66.80%	-	Yes	Source
Microsoft	Phi 4 Mini	01 Feb 2025	67%	-	Yes	Source
Microsoft	Phi 3 Mini 128K Instruct	-	68.50%	inferred family alias from phi-3.5-mini-instruct (score=0.3533; benches=31)	Yes	Source
Microsoft	Phi 3.5 mini instruct	23 Aug 2024	68.50%	-	Yes	Source
Google	Gemma 3n E4B	25 Jun 2025	71.70%	-	Yes	Source
Mistral	Mistral Nemo 12B	18 Jul 2024	76.80%	inferred family alias from mistral-nemo-instruct-2407 (score=0.3250; benches=8)	Yes	Source
Microsoft	Phi 3.5 MoE instruct	23 Aug 2024	81.30%	-	Yes	Source
Nvidia	Llama 3.1 Nemotron 70B Instruct	01 Oct 2024	84.53%	-	Yes	Source
Qwen	Qwen 2 Math 72B	-	85.10%	inferred high-confidence family alias from qwen2-72b-instruct (score=0.4667; benches=17)	Yes	Source
Qwen	Qwen 2 Math RM 72B	-	85.10%	inferred family alias from qwen2-72b-instruct (score=0.3917; benches=17)	Yes	Source
OpenAI	GPT 4 32K 0613	13 Jun 2023	87.50%	inferred high-confidence family alias from gpt-4-0613 (score=0.4899; benches=12)	Yes	Source