🧠 Reasoning League

#	Model	AIM	Năng lực
1	Claude Fable 5 Anthropic	94.7	100	NEW
2	Gemini 3.1 Pro Google	90.4	95	NEW
3	GPT-5.4 Pro OpenAI	89.7	94	NEW
4	GPT-5.5 OpenAI	89.5	94	NEW
5	Claude Opus 4.8 Anthropic	88.4	93	NEW
6	Claude Opus 4.7 Anthropic	86.6	91	NEW
7	GPT-5.4 OpenAI	84.8	89	NEW
8	GPT-5.5 Pro OpenAI	84.4	88	NEW
9	GLM-5.2 Zhipu AI	80.3	84	NEW
10	Gemini 3.1 Pro Preview Google	80.3	83	NEW
11	Gemini 3.5 Flash Google	76.1	79	NEW
12	Claude Opus 4.6 Anthropic	75.3	78	NEW
13	GPT-5.2 Pro OpenAI	74.3	77	NEW
14	Qwen3.7 Max Alibaba	73.9	76	NEW
15	MiMo-V2.5-Pro Xiaomi	71.5	73	NEW
16	Claude Sonnet 4.6 Anthropic	69.4	71	NEW
17	GLM-5.1 Zhipu AI	66.6	68	NEW
18	DeepSeek V4 Pro DeepSeek	65.8	67	NEW
19	GLM-5 Zhipu AI	64.8	66	NEW
20	Kimi K2.6 Moonshot AI	64.6	65	NEW
21	Qwen3.7 Plus Alibaba	64.1	65	NEW
22	Grok 4.20 xAI	63.6	64	NEW
23	GPT-5.4 mini OpenAI	62.4	63	NEW
24	GPT-5.2 OpenAI	61.1	61	NEW
25	Hy3 Preview Tencent	60.5	61	NEW
26	MiniMax M2.7 MiniMax	59.5	59	NEW
27	DeepSeek V4 Flash DeepSeek	58.9	59	NEW
28	Gemini 3 Pro Preview Google	56.8	56	NEW
29	Kimi K2.5 Moonshot AI	54.4	54	NEW
30	Grok 4.3 xAI	53.3	52	NEW
31	Gemini 3 Pro Deep Think Google	51.5	50	NEW
32	Nemotron 3 Ultra NVIDIA	50.6	49	NEW
33	GPT-5 Pro OpenAI	48.6	47	NEW
34	Claude Opus 4.5 Anthropic	45.4	43	NEW
35	o3 OpenAI	41.9	39	NEW
36	Gemma 4 31B Google	41.4	39	NEW
37	GPT-5 OpenAI	39.7	37	NEW
38	Gemini 3 Flash Google	39.1	36	NEW
39	GPT-5.1 OpenAI	37.4	34	NEW
40	Gemini 3 Pro Google	36.3	33	NEW
41	GPT-5 mini OpenAI	31.5	27	NEW
42	o4-mini OpenAI	29.5	25	NEW
43	Gemini 2.5 Flash Google	21.1	15	NEW
44	Claude Sonnet 4.5 Anthropic	20.4	14	NEW
45	Claude Opus 4.1 Anthropic	20.3	14	NEW
46	Grok 4 xAI	19.9	14	NEW
47	Gemini 2.5 Pro Google	19.3	13	NEW
48	Claude Opus 4 Anthropic	15.6	9	NEW
49	o1 Pro OpenAI	15.6	9	NEW
50	o1 OpenAI	15.4	9	NEW
51	Claude Sonnet 4 Anthropic	15.1	8	NEW
52	Llama 4 Maverick Meta	12.2	5	NEW
53	Claude 3.5 Sonnet Anthropic	9.9	2	NEW
54	Qwen3-235B-A22B-Instruct Alibaba	8.0	0	NEW
55	GPT-4o OpenAI	8.0	0	NEW
56	Trinity-Large-Thinking Arcee AI	8.0	0	NEW

Cách tính điểm — AIM Index

● Năng lực 65%● Phổ biến (mạng xã hội) 25%● Độ mới 10%

GPQA DiamondARC-AGIHLE

Nguồn dữ liệu thật (4)

benchlm.ai ↗llm-stats.com ↗labs.scale.com ↗artificialanalysis.ai ↗

Điểm minh hoạ (seed). Pipeline R3/R4 sẽ thay bằng AIM Index thật từ benchmark + mạng xã hội.