Wino grande — Epoch AI

Wino grande — Epoch AI is run by Epoch AI. It has ranked 137 models, of which the catalogue holds 43, scoring from 51.5 to 89.2 on Accuracy.

measured by Epoch AI · the board itself

Place	Model	Metric	Score
1stof 137	Llama 3.1 405B	Accuracy	89.2
2ndof 137	Claude 3 Opus	Accuracy	88.5
5thof 137	falcon-180B	Accuracy	87.1
6thof 137	DeepSeek-V2	Accuracy	86.3
7thof 137	DeepSeek-V3	Accuracy	85.2
12thof 137	Llama 3 70B	Accuracy	83.5
15thof 137	Meta-Llama-3.1-405B-Instruct	Accuracy	82.2
17thof 137	GPT-3.5 Turbo (older v0613)	Accuracy	81.6
18thof 137	Phi-3-small-8k-instruct	Accuracy	81.5
19thof 137	Phi-3-medium-128k-instruct	Accuracy	81.5
22ndof 137	Qwen 2.5 Coder 32B	Accuracy	80.8
23rdof 137	Llama-2-70b-hf	Accuracy	80.2
26thof 137	gemma-7b	Accuracy	79
28thof 137	falcon-11B	Accuracy	78.3
41stof 137	falcon-40b	Accuracy	76.9
49thof 137	Llama 3 8B	Accuracy	75.7
50thof 137	Mistral-7B-v0.1	Accuracy	75.3
52ndof 137	Claude 3 Sonnet	Accuracy	75.1
55thof 137	Claude 3 Haiku	Accuracy	74.2
60thof 137	phi-1_5	Accuracy	73.4
67thof 137	llama-13b	Accuracy	73
70thof 137	Yi-9B	Accuracy	73
72ndof 137	Qwen2.5-Coder-7B	Accuracy	72.9
74thof 137	Llama-2-13b	Accuracy	72.8
78thof 137	Yi-6B	Accuracy	71.3
81stof 137	Phi-3-mini-4k-instruct	Accuracy	70.8
89thof 137	llama-7b	Accuracy	70.1
96thof 137	Llama-2-7b	Accuracy	69.2
99thof 137	GPT-3.5 Turbo (1106)	Accuracy	68.8
104thof 137	falcon-7b	Accuracy	67.2
111thof 137	gpt-neox-20b	Accuracy	66.1
113thof 137	gemma-2b	Accuracy	65.4
114thof 137	Llama 3 8B Instruct	Accuracy	65
115thof 137	xgen-7b-8k-base	Accuracy	64.9
116thof 137	opt-1.3b	Accuracy	64.7
117thof 137	gpt-j-6b	Accuracy	64.5
118thof 137	starcoder2-15b	Accuracy	64.3
127thof 137	gpt2-xl	Accuracy	58.3
130thof 137	starcoder2-3b	Accuracy	57.1
131stof 137	starcoder2-7b	Accuracy	57.1
133rdof 137	Qwen2.5-Coder-0.5B	Accuracy	54.8
134thof 137	phi-2	Accuracy	54.7
137thof 137	stablelm-tuned-alpha-7b	Accuracy	51.5

Place

Model

Metric

Score

1stof 137

Llama 3.1 405B

Accuracy

89.2

2ndof 137

Claude 3 Opus

Accuracy

88.5