ALE-Bench · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	GPT-6 Astra	max	3164	–	–
2	GPT-6.1 Sol	max	2990	–	–
3	Fable 5.1	max	2849	–	–
4	Opus 5.5	max	2748	–	–
5	GPT-6 Sol	max	2637	–	–
6	Opus 5	max	2622	–	–
7	Fable 5	max	2463	–	–
8	Sonnet 5.5	max	2448	–	–
9	GPT-5.6 Sol	max	2410	–	–
10	Gemini 3.8 Flash	high	2182	–	–
11	GPT-5.6 Terra	max	2154	–	–
12	GPT-5.5	xhigh	2108	–	–
13	Opus 4.8	max	2102	–	–
14	Gemini 3.1 Pro	high	2041	–	–
15	Kimi K3	max	1991	–	–
16	GLM 5.3	high	1937	–	–
17	GPT-5.3 Codex	xhigh	1934	–	–
18	Opus 4.7	max	1897	–	–
19	Gemini 3.7 Flash	high	1876	–	–
20	DeepSeek V4.1 Flash	max	1829	–	–
21	GPT-5.6 Luna	max	1829	–	–
22	Gemini 3.6 Flash	high	1825	–	–
23	Grok 4.6	xhigh	1822	–	–
24	GPT-5.4	high	1811	–	–
25	GPT-6 Luna	xhigh	1791	–	–
26	Sonnet 5	max	1783	–	–
27	Sonnet 4.6	max	1768	–	–
28	DeepSeek V4 Pro 0813	max	1730	–	–
29	Gemini 3 Pro	high	1729	–	–
30	Gemini 3 Flash	high	1722	–	–
31	GPT-5.2	high	1715	–	–
32	Gemini 3.5 Flash	high	1712	–	–
33	Kimi K2.6	max	1695	–	–
34	DeepSeek V4 Flash 0731	max	1686	–	–
35	GPT-5.2 Codex	xhigh	1686	–	–
36	GLM 5.2	max	1685	–	–
37	Mistral Large 4	high	1657	–	–
38	GLM 5.1	max	1646	–	–
39	Opus 4.6	max	1607	–	–
40	Qwen 3.7 Max	max	1548	–	–
41	GPT-5	high	1545	–	–
42	GPT-5.1 Codex	xhigh	1538	–	–
43	Hy3	high	1516	–	–
44	DeepSeek V4 Pro	high	1489	–	–
45	GPT-5.4 Mini	high	1487	–	–
46	MiMo V2.6 Pro	high	1476	–	–
47	Grok 4.5	high	1467	–	–
48	Opus 4.5	high	1429	–	–
49	GPT-5.1	high	1425	–	–
50	Kimi K2.7 Code	max	1422	–	–
51	GPT-5 Mini	high	1421	–	–
52	Gemma 4 31B IT	high	1394	–	–
53	Grok 4.20	xhigh	1388	–	–
54	DeepSeek V4 Flash	max	1380	–	–
55	MiMo V2 Pro	–	1347	–	–
56	GLM 5	high	1313	–	–
57	GPT-5.4 Nano	high	1312	–	–
58	o3	high	1304	–	–
59	GLM 5.3 Flash	high	1283	–	–
60	Kimi K2.5	max	1279	–	–
61	Qwen 3.6 Plus	max	1273	–	–
62	o4 mini	high	1271	–	–
63	MiMo V2.5 Pro	xhigh	1254	–	–
64	Grok 4.3	xhigh	1211	–	–
65	Inkling	max	1195	–	–
66	Sonnet 4.5	high	1192	–	–
67	Mercury 2	high	1190	–	–
68	Opus 4.1	high	1183	–	–
69	Gemini 2.5 Pro	high	1175	–	–
70	Gemini 3.1 Flash-Lite	high	1165	–	–
71	Gemma 4 26B	xhigh	1165	–	–
72	Gemini 3.5 Flash-Lite	high	1137	–	–
73	Mistral Medium 3.5	high	1131	–	–
74	gpt-oss-120b	high	1119	–	–
75	Grok 4.1 Fast	high	1111	–	–
76	MiMo V2.5	xhigh	1106	–	–
77	Kimi K2	medium	1103	–	–
78	Sonnet 4	high	1088	–	–
79	DeepSeek V3.1	none	1078	–	–
80	Qwen 3.5 Plus	none	1071	–	–
81	DeepSeek R1	xhigh	1059	–	–
82	MiniMax M3	max	1053	–	–
83	MiniMax M2.5	high	1033	–	–
84	Gemini 2.5 Flash	medium	1022	–	–
85	MiniMax M2.7	high	1021	–	–
86	Nemotron 3.5 Lightning	–	1017	–	–
87	GPT-5 Nano	high	1016	–	–
88	Qwen 3.6 Flash	max	1015	–	–
89	GLM 4.7	none	1011	–	–
90	Qwen 3.5 397B-A17B	max	1007	–	–
91	Haiku 4.5	xhigh	1006	–	–
92	MiMo V2 Flash	high	1000	–	–
93	MiniMax M2.1	–	998	–	–
94	GPT-4.1	high	985	–	–
95	Mercury 2.5	high	977	–	–
96	Qwen 3.7 Flash	max	946	–	–
97	gpt-oss-20b	high	942	–	–
98	Qwen 3 Max	max	894	–	–
99	Qwen 3.5 27B	high	882	–	–
100	Mistral Small 4	high	862	–	–
101	GLM 4.6	none	838	–	–
102	Qwen 3 Coder	medium	827	–	–
103	Qwen 3.5 Flash	max	826	–	–
104	Mistral Large 3.2512	–	721	–	–
105	Mistral Medium 3.1	none	701	–	–
106	Qwen 3.5 35B-A3B	low	648	–	–
107	GLM 4.5	xhigh	552	–	–
108	Llama 4 Maverick	medium	552	–	–
109	Gemini 2.5 Flash Lite	none	542	–	–
110	Codestral 2508	–	341	–	–