GPQA Diamond (Epoch AI run)

Graduate-level science multiple choice (the diamond subset), run by Epoch AI. Accuracy. Data: Epoch AI, CC BY.

164models

–Open Index weight

–last updated

epoch-taskspublisher

#	Model	Effort	Score	Cost / task	Harness
1	Gemini 4	high	96.4%	–	Epoch AI
2	GPT-6 Astra	max	95.8%	–	Epoch AI
3	Sonnet 5.5	max	95.6%	–	Epoch AI
4	GPT-6.1 Sol	max	95.4%	–	Epoch AI
5	Gemini 3.8 Flash	high	95.4%	–	Epoch AI
6	Gemini 3.7 Flash	high	94.8%	–	Epoch AI
7	GPT-5.4 Pro	xhigh	94.6%	–	Epoch AI
8	Gemini 3.1 Pro	high	94.4%	–	Epoch AI
9	GPT-6 Sol	max	94.3%	–	Epoch AI
10	Gemini 3.6 Flash	high	94.1%	–	Epoch AI
11	Grok 4.6	high	94.0%	–	Epoch AI
12	GPT-5.5	xhigh	94.0%	–	Epoch AI
13	GPT-5.5 Pro	xhigh	93.9%	–	Epoch AI
14	GPT-5.6 Sol	max	93.5%	–	Epoch AI
15	Grok 4.5	high	93.4%	–	Epoch AI
16	GPT-5.6 Terra	max	93.3%	–	Epoch AI
17	GPT-5.4	xhigh	93.3%	–	Epoch AI
18	Kimi K3	max	93.1%	–	Epoch AI
19	Opus 5	max	92.9%	–	Epoch AI
20	Gemini 3.5 Flash	high	92.8%	–	Epoch AI
21	Grok 4.7	xhigh	92.7%	–	Epoch AI
22	Gemini 3 Pro	high	92.6%	–	Epoch AI
23	Qwen 3.8 Max	xhigh	92.3%	–	Epoch AI
24	GLM 5.2	max	91.9%	–	Epoch AI
25	DeepSeek V4 Pro 0813	max	91.7%	–	Epoch AI
26	GPT-5.6 Luna	max	91.6%	–	Epoch AI
27	GPT-5.2	xhigh	91.4%	–	Epoch AI
28	Opus 4.8	max	91.0%	–	Epoch AI
29	DeepSeek V4 Flash 0731	max	91.0%	–	Epoch AI
30	GLM 5.3	max	90.9%	–	Epoch AI
31	MiniMax M3	max	90.9%	–	Epoch AI
32	Qwen 3.7 Max	max	90.9%	–	Epoch AI
33	DeepSeek V4 Pro	high	90.9%	–	Epoch AI
34	Kimi K2.6	max	90.8%	–	Epoch AI
35	Opus 5.5	max	90.6%	–	Epoch AI
36	Sonnet 5	xhigh	90.5%	–	Epoch AI
37	GPT-6 Luna	max	90.5%	–	Epoch AI
38	GLM 5.3 Flash	max	90.2%	–	Epoch AI
39	Opus 4.7	xhigh	90.2%	–	Epoch AI
40	GLM 5.1	max	89.9%	–	Epoch AI
41	DeepSeek V4.1 Flash	max	89.8%	–	Epoch AI
42	Muse Spark	–	89.8%	–	Epoch AI
43	Haiku 5.5	max	89.6%	–	Epoch AI
44	Gemini 3 Flash	high	89.4%	–	Epoch AI
45	Grok 4.20	xhigh	89.3%	–	Epoch AI
46	Grok 4.3	high	88.8%	–	Epoch AI
47	Inkling Small	xhigh	88.5%	–	Epoch AI
48	Qwen 3.6 Plus	max	88.4%	–	Epoch AI
49	Opus 4.6	max	88.4%	–	Epoch AI
50	Inkling	xhigh	88.3%	–	Epoch AI
51	Mistral Large 4	high	88.1%	–	Epoch AI
52	Kimi K2.7 Code	max	87.9%	–	Epoch AI
53	Qwen 3.7 Plus	high	87.9%	–	Epoch AI
54	GLM 5	high	87.8%	–	Epoch AI
55	GPT-5.1	high	87.6%	–	Epoch AI
56	Kimi K2.5	max	87.6%	–	Epoch AI
57	Qwen 3.6 Max	max	87.4%	–	Epoch AI
58	Grok 4	high	87.0%	–	Epoch AI
59	GPT-5.4 Mini	xhigh	86.9%	–	Epoch AI
60	Qwen 3.5 397B-A17B	none	86.4%	–	Epoch AI
61	GPT-5	high	86.2%	–	Epoch AI
62	Qwen 3.6 27B	max	85.9%	–	Epoch AI
63	Fable 5	max	85.9%	–	Epoch AI
64	Opus 4.5	high	85.5%	–	Epoch AI
65	Nemotron 3 Ultra 550B A55B	max	85.4%	–	Epoch AI
66	Gemini 2.5 Pro	high	85.3%	–	Epoch AI
67	Qwen 3.5 Plus	none	84.8%	–	Epoch AI
68	Qwen 3.6 35B-A3B	none	84.8%	–	Epoch AI
69	Kimi K2	medium	84.2%	–	Epoch AI
70	Qwen 3.5 35B-A3B	low	83.5%	–	Epoch AI
71	Qwen 3.6 Flash	max	83.3%	–	Epoch AI
72	Gemini 3.5 Flash-Lite	high	83.3%	–	Epoch AI
73	Sonnet 4.6	high	83.3%	–	Epoch AI
74	GLM 4.7	none	83.3%	–	Epoch AI
75	GPT-5.5 Instant	–	82.5%	–	Epoch AI
76	Qwen 3.7 Flash	max	82.3%	–	Epoch AI
77	Qwen 3.5 Flash	max	82.3%	–	Epoch AI
78	o3	high	81.8%	–	Epoch AI
79	Gemini 3.1 Flash-Lite	high	81.8%	–	Epoch AI
80	Qwen 3 235B-A22B	xhigh	80.1%	–	Epoch AI
81	o4 mini	high	79.6%	–	Epoch AI
82	Sonnet 4	high	79.2%	–	Epoch AI
83	Qwen 3.5 9B	low	79.0%	–	Epoch AI
84	Sonnet 4.5	high	78.8%	–	Epoch AI
85	GPT-5.4 Nano	high	78.5%	–	Epoch AI
86	o3 mini	high	77.0%	–	Epoch AI
87	o1	high	76.8%	–	Epoch AI
88	Opus 4.1	high	76.8%	–	Epoch AI
89	Grok 3 Mini	low	76.3%	–	Epoch AI
90	Opus 4	medium	76.3%	–	Epoch AI
91	Gemma 4 31B IT	minimal	75.8%	–	Epoch AI
92	gpt-oss-120b	high	75.8%	–	Epoch AI
93	Grok 3	none	75.8%	–	Epoch AI
94	GPT-5 Mini	high	75.0%	–	Epoch AI
95	Gemma 4 26B	minimal	73.2%	–	Epoch AI
96	Qwen 3 Max	max	72.6%	–	Epoch AI
97	DeepSeek V3.2	high	71.2%	–	Epoch AI
98	Haiku 4.5	xhigh	71.2%	–	Epoch AI
99	Qwen 3	high	70.7%	–	Epoch AI
100	Qwen 3 30B-A3B	none	70.1%	–	Epoch AI
101	GPT-5 Nano	high	69.4%	–	Epoch AI
102	GPT-4.5	none	68.7%	–	Epoch AI
103	DeepSeek V3	none	67.6%	–	Epoch AI
104	Llama 4 Maverick	medium	67.0%	–	Epoch AI
105	GPT-4.1	high	66.9%	–	Epoch AI
106	Sonnet 3.7	medium	66.0%	–	Epoch AI
107	GPT-4.1 Mini	high	65.8%	–	Epoch AI
108	Gemini 2.0 Pro	none	65.7%	–	Epoch AI
109	DeepSeek R1	xhigh	64.1%	–	Epoch AI
110	o1 mini	high	62.4%	–	Epoch AI
111	gpt-oss-20b	medium	60.8%	–	Epoch AI
112	Mistral Medium 3	none	59.5%	–	Epoch AI
113	Gemini 2.0 Flash	medium	57.1%	–	Epoch AI
114	Qwen 2.5 Max	none	56.1%	–	Epoch AI
115	Magistral Small 1.0	–	56.1%	–	Epoch AI
116	Qwen 3 32B	none	54.1%	–	Epoch AI
117	Grok 2	none	53.8%	–	Epoch AI
118	Qwen 3 14B	none	53.4%	–	Epoch AI
119	Llama 4 Scout	medium	51.8%	–	Epoch AI
120	Mistral Large 2	none	51.3%	–	Epoch AI
121	Llama 3.1 405b	none	50.9%	–	Epoch AI
122	Qwen 2.5 72B	none	49.1%	–	Epoch AI
123	Mistral Small 3.2	none	49.1%	–	Epoch AI
124	GPT-4.1 Nano	high	48.9%	–	Epoch AI
125	GPT-4o	high	47.9%	–	Epoch AI
126	Gemma 3 27B	none	47.7%	–	Epoch AI
127	Magistral Small 1.2	–	47.6%	–	Epoch AI
128	Qwen 3 8B	none	47.5%	–	Epoch AI
129	Llama 3.3 70b	none	47.4%	–	Epoch AI
130	Mistral Small 3	none	47.3%	–	Epoch AI
131	Opus 3	none	47.2%	–	Epoch AI
132	Qwen 2.5 32B	none	46.1%	–	Epoch AI
133	Gemini 1.5 Pro	none	45.9%	–	Epoch AI
134	Qwen 3 4B	none	45.8%	–	Epoch AI
135	GLM 4.7 Flash	none	45.1%	–	Epoch AI
136	Llama 3.1 70b	none	44.2%	–	Epoch AI
137	Mistral Small 3.1	none	41.9%	–	Epoch AI
138	Llama 3.2 90b	none	41.0%	–	Epoch AI
139	Qwen 2 72B	none	40.8%	–	Epoch AI
140	Sonnet 3	none	40.6%	–	Epoch AI
141	Gemma 3 12B	none	39.5%	–	Epoch AI
142	Mistral Large	none	38.8%	–	Epoch AI
143	GPT-4o mini	high	37.7%	–	Epoch AI
144	Llama 3 70b	none	37.5%	–	Epoch AI
145	Gemma 2 27B	–	36.5%	–	Epoch AI
146	Haiku 3	medium	36.3%	–	Epoch AI
147	GPT-4	high	35.7%	–	Epoch AI
148	Qwen 2.5 7B	–	35.5%	–	Epoch AI
149	Gemini 1.0 Pro	none	34.0%	–	Epoch AI
150	Gemini 1.5 Flash	none	33.0%	–	Epoch AI
151	Qwen 1.5 32B	–	30.7%	–	Epoch AI
152	Qwen 3.1	none	30.6%	–	Epoch AI
153	Mixtral 8x7B Instruct	none	30.6%	–	Epoch AI
154	Qwen 1.5 72B	–	28.8%	–	Epoch AI
155	GPT-3.5	high	28.0%	–	Epoch AI
156	Gemma 2 9B	–	27.5%	–	Epoch AI
157	Llama 3.1 8b	medium	27.0%	–	Epoch AI
158	Llama 2 70b	none	26.3%	–	Epoch AI
159	Llama 3 8b	none	26.1%	–	Epoch AI
160	Llama 3.2 1b	none	23.9%	–	Epoch AI
161	Gemma 3 4B	none	23.2%	–	Epoch AI
162	Gemma 3 1B	none	19.9%	–	Epoch AI
163	Mistral	none	15.2%	–	Epoch AI
164	DeepSeek R1.0528	–	9.3%	–	Epoch AI