Best AI models for critical thinking (Watson-Glaser)

Ranked by accuracy on the meo “critical thinking (Watson-Glaser)” domain — one of eleven un-leaked, contamination-resistant domains. Per-domain strength reveals where a model genuinely reasons versus where it pattern-matches; the leaders differ sharply by domain.

Ranking 82 models across the meo-tested roster · as of 2026-10-18.

#	Model	Lab	accuracy
1	OpenAI: GPT-5.5	openai	100%
2	Google: Gemini 4 Argon	google	100%
3	Anthropic: Claude Opus 5.5	anthropic	100%
4	OpenAI: GPT-6.1 Sol	openai	100%
5	Anthropic: Claude Fable 5	anthropic	100%
6	Anthropic: Claude Opus 4.8	anthropic	100%
7	Google: Gemini 3.5 Flash	google	100%
8	MoonshotAI: Kimi K3	moonshotai	100%
9	Google: Gemini 3.1 Pro Preview	google	100%
10	Qwen: Qwen3.7 Plus	qwen	100%
11	Nex AGI: Nex-N2-Pro (free)	nex-agi	100%
12	Z.ai: GLM 4.6V	z-ai	100%
13	Google: Gemini 3.1 Flash Lite Preview	google	100%
14	Mistral: Mistral Large 3 2512	mistralai	100%
15	Google: Gemini 3.1 Flash Lite	google	100%
16	Meta: Llama 4 Maverick	meta-llama	100%
17	Google: Gemma 3 12B	google	100%
18	Qwen: Qwen3.6 Plus	qwen	100%
19	OpenAI: GPT-5.5 Pro	openai	75%
20	Z.ai: GLM 5.2	z-ai	75%
21	Qwen: Qwen3.7 Max	qwen	75%
22	OpenAI: GPT-5.4 Nano	openai	75%
23	Qwen: Qwen3.5-Flash	qwen	75%
24	Qwen: Qwen3 VL 30B A3B Instruct	qwen	75%
25	Qwen: Qwen3 VL 30B A3B Thinking	qwen	75%
26	Google: Gemma 4 31B	google	75%
27	OpenAI: GPT-4.1 Mini	openai	75%
28	Baidu: ERNIE 4.5 VL 424B A47B	baidu	75%
29	Qwen: Qwen3.5 Plus 2026-02-15	qwen	75%
30	Google: Gemini 2.5 Flash Lite	google	75%
31	Google: Gemma 4 26B A4B	google	75%
32	Qwen: Qwen3.5-27B	qwen	75%
33	OpenAI: GPT-4.1 Nano	openai	75%
34	ByteDance Seed: Seed-2.0-Lite	bytedance-seed	75%
35	Qwen: Qwen3.5-35B-A3B	qwen	75%
36	Owl Alpha	openrouter	50%
37	Xiaomi: MiMo-V2.5-Pro	xiaomi	50%
38	Qwen: Qwen3.6 Flash	qwen	50%
39	MoonshotAI: Kimi K2.7 Code	moonshotai	50%
40	OpenAI: GPT-5 Mini	openai	50%
41	Qwen: Qwen3 VL 32B Instruct	qwen	50%
42	Z.ai: GLM 5.1	z-ai	50%
43	ByteDance Seed: Seed 1.6 Flash	bytedance-seed	50%
44	Mistral: Mistral Small 4	mistralai	50%
45	Perceptron: Perceptron Mk1	perceptron	50%
46	Mistral: Ministral 3 8B 2512	mistralai	50%
47	Mistral: Mistral Small 3.2 24B	mistralai	50%
48	Qwen: Qwen2.5 VL 72B Instruct	qwen	50%
49	MiniMax: MiniMax-01	minimax	50%
50	OpenAI: GPT-4o-mini	openai	50%
51	Qwen: Qwen3.6 35B A3B	qwen	50%
52	Google: Gemma 3 4B	google	50%
53	Perplexity: Sonar	perplexity	50%
54	NVIDIA: Nemotron Nano 12B 2 VL (free)	nvidia	50%
55	Anthropic: Claude 3 Haiku	anthropic	50%
56	inclusionAI: Ring-2.6-1T	inclusionai	25%
57	DeepSeek: DeepSeek V4 Flash 0423	deepseek	25%
58	DeepSeek: DeepSeek V4 Pro 0423	deepseek	25%
59	SpaceXAI: Grok 4.7	x-ai	25%
60	MiniMax: MiniMax M3	minimax	25%
61	MoonshotAI: Kimi K2.6	moonshotai	25%
62	OpenAI: GPT-5.1-Codex-Mini	openai	25%
63	ByteDance Seed: Seed 1.6	bytedance-seed	25%
64	Qwen: Qwen3 VL 235B A22B Instruct	qwen	25%
65	Qwen: Qwen3 VL 8B Instruct	qwen	25%
66	Tencent: Hy3 preview	tencent	25%
67	Google: Gemma 3 27B	google	25%
68	StepFun: Step 3.7 Flash	stepfun	25%
69	Meta: Llama 4 Scout	meta-llama	25%
70	Mistral: Ministral 3 3B 2512	mistralai	25%
71	Arcee AI: Trinity Large Thinking	arcee-ai	25%
72	Amazon: Nova Lite 1.0	amazon	25%
73	Meta: Llama 3.2 11B Vision Instruct	meta-llama	25%
74	NVIDIA: Nemotron 3 Ultra	nvidia	0%
75	SpaceXAI: Grok 4.3	x-ai	0%
76	Xiaomi: MiMo-V2.5	xiaomi	0%
77	Qwen: Qwen3 VL 8B Thinking	qwen	0%
78	Mistral: Ministral 3 14B 2512	mistralai	0%
79	OpenAI: GPT-5 Nano	openai	0%
80	NVIDIA: Nemotron 3 Nano Omni (free)	nvidia	0%
81	Qwen: Qwen3.5-9B	qwen	0%
82	ByteDance Seed: Seed-2.0-Mini	bytedance-seed	0%

← All rankings Methodology & 𝕍 →