EQ-Bench Longform Creative Writing

#	Model	Effort	Score	Cost / task	Harness
1	Opus 5	max	86.3%	–	–
2	Fable 5.1	max	85.3%	–	–
3	Opus 5.5	max	84.6%	–	–
4	Fable 5	max	83.0%	–	–
5	Grok 4.7	max	83.0%	–	–
6	GPT-6 Astra	max	82.8%	–	–
7	Muse Spark 1.3	max	82.8%	–	–
8	GPT-6 Sol	max	82.8%	–	–
9	Opus 4.7	max	81.8%	–	–
10	GLM 5.3	max	81.8%	–	–
11	GPT-5.6 Sol	max	81.7%	–	–
12	Muse Spark 1.2	max	81.5%	–	–
13	Opus 4.8	max	80.8%	–	–
14	Sonnet 4.6	max	79.9%	–	–
15	Kimi K3	max	79.6%	–	–
16	Kimi K2.6	max	78.5%	–	–
17	GPT-5.4	xhigh	78.3%	–	–
18	Sonnet 5	max	78.3%	–	–
19	GPT-5.5	xhigh	78.2%	–	–
20	GPT-5.6 Terra	max	78.0%	–	–
21	GLM 5.2	max	77.9%	–	–
22	Opus 4.6	max	77.7%	–	–
23	Gemini 3.8 Flash	max	76.8%	–	–
24	DeepSeek V4 Pro	max	75.6%	–	–
25	GPT-5.6 Luna	max	75.5%	–	–
26	Kimi K2.5	max	74.9%	–	–
27	DeepSeek V4.1 Flash	max	74.0%	–	–
28	MiMo V2.5 Pro	xhigh	73.7%	–	–
29	GLM 5.1	max	73.5%	–	–
30	Opus 4.5	high	73.1%	–	–
31	GPT-5.2025	high	73.1%	–	–
32	GPT-5.4 Mini	xhigh	73.0%	–	–
33	Inkling	max	72.5%	–	–
34	Gemini 3.5 Flash	xhigh	71.8%	–	–
35	GPT-5.2	xhigh	71.6%	–	–
36	Sonnet 4.5	high	71.4%	–	–
37	GLM 5	high	70.9%	–	–
38	MiMo V2 Pro	–	70.9%	–	–
39	Muse Spark 1.1	max	70.6%	–	–
40	Opus 4.1	high	70.5%	–	–
41	Kimi K2	medium	69.0%	–	–
42	Gemini 3 Pro	high	68.8%	–	–
43	Grok 4.20	xhigh	68.5%	–	–
44	Qwen 3.5 397B-A17B	max	68.3%	–	–
45	Gemini 3.1 Pro	high	68.2%	–	–
46	DeepSeek V3.2	high	66.8%	–	–
47	GPT-5	high	66.8%	–	–
48	Qwen 3 Max	max	66.2%	–	–
49	GLM 4.7	none	66.0%	–	–
50	DeepSeek V4 Flash	max	66.0%	–	–
51	MiniMax M2.5	high	65.3%	–	–
52	Haiku 4.5	xhigh	65.0%	–	–
53	o3	high	62.7%	–	–
54	GPT-5.3	xhigh	62.6%	–	–
55	DeepSeek V3.1	none	62.3%	–	–
56	DeepSeek V4 Flash 0731	max	61.2%	–	–
57	DeepSeek V3.0324	–	60.1%	–	–
58	DeepSeek R1	xhigh	59.5%	–	–
59	DeepSeek R1.0528	–	59.1%	–	–
60	Qwen 3.5 27B	high	59.0%	–	–
61	Gemini 2.5 Pro	high	58.8%	–	–
62	Sonnet 4	high	58.4%	–	–
63	Sonnet 3.5	medium	58.4%	–	–
64	MiniMax M2.1	–	57.8%	–	–
65	GPT-5.4 Nano	xhigh	57.8%	–	–
66	Sonnet 3.7	medium	57.7%	–	–
67	Qwen 3 235B-A22B	xhigh	57.4%	–	–
68	GLM 4.6	none	57.3%	–	–
69	Gemma 4 31B IT	high	56.5%	–	–
70	GPT-5 Mini	high	56.2%	–	–
71	Gemini 3 Flash	high	55.7%	–	–
72	GLM 4.5	xhigh	55.5%	–	–
73	Qwen 3.5 122B-A10B	none	54.3%	–	–
74	Qwen 3.8 27B	max	53.3%	–	–
75	MiMo V2 Flash	high	53.1%	–	–
76	Gemma 4 26B	xhigh	50.7%	–	–
77	Gemini 3.1 Flash-Lite	high	49.0%	–	–
78	Gemma 4 12B IT	–	48.7%	–	–
79	Gemini 2.5 Flash	medium	48.6%	–	–
80	Mistral Large 3	none	48.3%	–	–
81	GLM 4.7 Flash	none	47.8%	–	–
82	MiniMax M2	–	46.4%	–	–
83	Mistral Medium 3.1	none	46.2%	–	–
84	Grok 4	high	45.7%	–	–
85	Gemma 3 27B IT	medium	45.2%	–	–
86	Qwen 3.5 35B-A3B	low	44.5%	–	–
87	Mistral Small 4	high	43.2%	–	–
88	Gemma 3 12B IT	–	41.8%	–	–
89	Mistral Small 3.2	none	41.6%	–	–
90	Qwen 3 32B	none	40.5%	–	–
91	Grok 4.1 Fast	high	40.1%	–	–
92	Qwen 3 14B	none	35.9%	–	–
93	gpt-oss-120b	high	35.8%	–	–
94	Qwen 3 30B-A3B	none	35.6%	–	–
95	Gemma 3 4B IT	–	34.4%	–	–
96	Mistral Small	none	32.6%	–	–
97	GPT-4.1 Nano	high	32.4%	–	–
98	Llama 4 Maverick	medium	31.9%	–	–
99	Qwen 3	high	31.5%	–	–
100	Llama 4 Scout	medium	31.1%	–	–
101	Mistral	none	28.7%	–	–
102	GPT-5 Nano	high	28.0%	–	–
103	gpt-oss-20b	high	21.2%	–	–
104	Qwen 3 4B	none	17.0%	–	–