MLCR-AA · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	Sonnet 5.5	max	75.0%	–	–
2	Fable 5.1	max	71.1%	–	–
3	Opus 5.5	max	66.7%	–	–
4	Fable 5	max	64.4%	–	–
5	Haiku 5.5	max	63.3%	–	–
6	Opus 5	high	59.4%	–	–
7	Sonnet 5	max	55.0%	–	–
8	GLM 5.3 Flash	max	51.1%	–	–
9	GLM 5.3	max	48.3%	–	–
10	Opus 4.8	max	45.6%	–	–
11	Muse Spark 1.3	max	43.3%	–	–
12	Kimi K3	max	38.3%	–	–
13	GPT-6 Astra	max	35.0%	–	–
14	GPT-6.1 Sol	max	33.9%	–	–
15	GPT-5.6 Terra	max	31.7%	–	–
16	Muse Spark 1.2	xhigh	31.1%	–	–
17	Opus 4.7	high	28.9%	–	–
18	GPT-5.6 Sol	max	26.1%	–	–
19	Sonnet 4.6	max	24.4%	–	–
20	DeepSeek V4.1 Flash	max	22.8%	–	–
21	Gemini 3.8 Flash	high	21.7%	–	–
22	Qwen 3.8 27B	xhigh	21.7%	–	–
23	DeepSeek V4 Pro 0424	max	21.1%	–	–
24	Qwen 3.8 Max	max	20.0%	–	–
25	Muse Glimmer (High)	high	20.0%	–	–
26	GPT-5.6 Luna	max	19.4%	–	–
27	MiMo V2.6 Pro	–	18.3%	–	–
28	Gemini 3.5 Flash	high	18.3%	–	–
29	GPT-5.5	xhigh	17.8%	–	–
30	DeepSeek V4 Pro 0813	max	17.8%	–	–
31	MiniMax M3	–	17.2%	–	–
32	Step 5 Preview	–	16.7%	–	–
33	GPT-6 Sol	max	16.1%	–	–
34	GPT-6 Luna	max	16.1%	–	–
35	Kimi K2.7 Code	–	16.1%	–	–
36	Gemini 3.1 Pro	–	15.6%	–	–
37	Grok 4.7	xhigh	15.0%	–	–
38	Gemini 3.7 Flash	high	15.0%	–	–
39	Grok 4.5	high	15.0%	–	–
40	DeepSeek V4 Flash 0420	max	15.0%	–	–
41	Gemini 3.6 Flash	high	14.4%	–	–
42	DeepSeek V4 Flash 0731	max	13.3%	–	–
43	Grok 4.6	high	12.2%	–	–
44	Inkling	xhigh	12.2%	–	–
45	Gemini 3 Flash	none	11.7%	–	–
46	Nemotron 3 Ultra	max	11.1%	–	–
47	Grok 4.3	high	10.0%	–	–
48	MiMo V2.5 Pro	–	9.4%	–	–
49	Qwen 3.7 Plus	–	9.4%	–	–
50	Gemini 2.5 Flash	–	8.9%	–	–
51	GLM 5.2	max	7.2%	–	–
52	Gemini 3.5 Flash-Lite	high	7.2%	–	–
53	K2 Horizon 375B A23B	–	6.7%	–	–
54	Haiku 4.5	–	6.1%	–	–
55	Gemini 3.1 Flash-Lite	high	6.1%	–	–
56	GPT-5.4 Nano	xhigh	5.0%	–	–
57	GPT-5.4 Mini	xhigh	3.9%	–	–
58	Nova 2.0 Pro Preview (Medium)	medium	3.9%	–	–
59	MiniMax M2.7	–	3.3%	–	–
60	Nemotron 3 Super 120B A12B (Reasoning)	–	3.3%	–	–
61	Llama 4 Maverick	none	2.2%	–	–
62	Gemini 2.5 Flash Lite	–	2.2%	–	–
63	Mistral Medium 3.5	high	1.7%	–	–
64	Nemotron 3.5 Lightning	–	1.1%	–	–
65	gpt-oss-120b	high	1.1%	–	–
66	Command A	–	0.6%	–	–
67	gpt-oss-20b	high	0.6%	–	–
68	Solar Pro 3	high	0.6%	–	–
69	Qwen 3.8 2.4T-A95B	–	0.0%	–	–
70	Nova Lite	none	0.0%	–	–
71	GPT-4o mini	none	0.0%	–	–