LynxModel

AI Model Intelligence RankingsAI 模型智能排行

62 frontier models, one transparent 100-point Intelligence Score across 11 public benchmarks (reasoning, coding, knowledge, math, agent, human preference). Every score links to its source. Data refreshed 2026-09-12. Click any column to sort.

62 个前沿模型,一套透明的百分制智能评分,覆盖 11 个公开 benchmark(推理/代码/知识/数学/智能体/人类偏好)。每个分数都带来源链接。数据更新于 2026-09-12。点表头可排序。

Find my model帮我选型 Pricing & value价格与性价比 Head-to-head模型对战 Image arena图片生成榜 Video arena视频生成榜 How the score works评分方法 Benchmark explorerBenchmark 百科

Rea = Reasoning推理 Cod = Coding代码 Kno = Knowledge知识 Mat = Math数学 Agt = Agent智能体 Pre = Preference偏好

⚖ Adjust the weights — the ranking recomputes live⚖ 调权重——排名实时重算

Reasoning推理30

Coding代码25

Knowledge知识15

Math数学15

Agent智能体10

Preference偏好5

Same formula as the published score: weighted average over the dimensions a model actually has, renormalized by the weights present. Below 30% coverage a model shows no score; below 60% it carries a low-data badge.与公开评分同一公式:按模型实际拥有的维度加权平均、按出席权重重归一;覆盖<30% 不出分,<60% 打"数据少"徽标。

Intelligence vs price — the value map智能分 vs 价格——性价比地图

Each dot is a scored model: higher = smarter, further left = cheaper. Top-left is the sweet spot. Color = provider, dashed lines = medians.每个点是一个有评分的模型:越往上越聪明,越往左越便宜,左上角是甜区。颜色=厂商,虚线=中位数。 Click to enlarge. 点击放大。

Intelligence vs price scatter — each dot a scored model, color = provider

#	Model模型	Score总分	Rea推理	Cod代码	Kno知识	Mat数学	Agt智能体	Pre偏好	Ctx上下文	$ in/1M输入$/1M	$ out/1M输出$/1M
1	Claude Fable 5(max)Anthropic	94.4	95.7	96.2	96.3	99.3	99.4	51.8	1M	$10	$50
2	GPT-5.6 Sol(max)OpenAI	93.4	90.4	96.3	93.8	98.4	92.9	81	1.1M	$2.00	$10
3	Qwen3.8 Max (0902)Qwen	90.4	97.4	93.2	—	81.5	—	48	1M	$2.00	$6.00
4	Qwen3.7 Max(on/off开/关)Qwen	88.5	95	85.7	94.3	85.7	95.6	43.8	1M	$1.48	$4.42
5	DeepSeek V4 Flash 0731DeepSeek	88.4	87.2	88.4	—	93.1	—	—	1.3M	$0.040	$0.080
6	Claude Opus 5(max)Anthropic	88	96.4	85.9	96.4	68.1	78.6	100	1M	$5.00	$25
7	GPT-5.6 Terra(max)OpenAI	87.9	88.8	88.8	—	96.3	87.1	55.8	1.1M	$2.00	$12
8	DeepSeek V4 Pro 0423(max)DeepSeek	86.3	89.2	85.9	92.1	87.5	97.1	39.8	1M	$1.60	$3.20
9	GLM 5.2(max)Z.ai	86.3	95.4	73	93.6	98.6	100	43	1M	$0.600	$2.00
10	Kimi K3(always-on常开)Moonshot AI	85.6	87.5	97.7	—	71.8	93.8	47.5	1M	$2.65	$13
11	GPT-5.6 Sol Pro(max)OpenAI	85.1 low data数据少	91.7	—	—	—	—	45.5	1.1M	$2.00	$10
12	Claude Opus 4.8(max)Anthropic	84.4	89.4	90	—	77.2	86.2	43.8	1M	$5.00	$25
13	Claude Sonnet 5(high)Anthropic	83.6	89.1	88.2	92.2	76.2	75.5	40.8	1M	$2.00	$10
14	GPT-5.6 Luna(max)OpenAI	83.3	80.1	91.9	—	94.5	—	38.2	1.1M	$0.200	$1.20
15	GPT-5.5(max)OpenAI	82.9	80.5	90.6	90.6	97.9	55.4	45.8	1.1M	$5.00	$30
16	Gemini 3.1 Pro Preview(high)Google	82.4	90.5	85.6	94.5	58.5	85.7	46.8	1M	$2.00	$12
17	DeepSeek V4 Flash 0423(max)DeepSeek	82	73.7	88.4	90.7	97	96.5	34	1M	$0.066	$0.131
18	Qwen3.7 Plus(on/off开/关)Qwen	81.7	77.5	86.9	93.2	85.7	93.8	39.8	1M	$0.320	$1.28
19	Qwen3.7 Flash(on/off开/关)Qwen	81.6 low data数据少	94.4	87.3	—	67.3	—	44	1M	$0.030	$0.130
20	Hy3Tencent	81.3 low data数据少	88.3	—	—	—	—	39.5	262K	$0.083	$0.330
21	Gemini 3.6 Flash(high)Google	81.1	84.5	91.7	58.3	100	—	46.3	1M	$0.750	$3.75
22	Grok 4.5(high)xAI	80.5	83.7	88.6	93.9	74.1	38	42.5	500K	$2.00	$6.00
23	Gemini 3.5 Flash(high)Google	80.1	83.3	85.1	—	72	96.2	44.5	1M	$1.50	$9.00
24	Grok 4.20(high)xAI	80.1	91.8	75.8	100	56.2	—	44	2M	$1.25	$2.50
25	MiniMax M3MiniMax	79.8	80.1	82	82.2	85.8	82.2	36.2	1M	$0.300	$1.20
26	Muse Spark 1.2meta	79.8 low data数据少	82.7	90.2	—	—	—	49.8	1M	$1.25	$4.25
27	Step 3.7 FlashStepFun	78.8	83.2	74.3	—	97.3	99.4	24	262K	$0.200	$1.15
28	Seed-2.0-MiniByteDance Seed	77.7 low data数据少	82.7	69.8	—	87	—	—	262K	$0.100	$0.400
29	Kimi K2.7 CodeMoonshot AI	76.5	69.3	82	—	100	90.9	40.5	262K	$0.710	$3.50
30	Nemotron 3 UltraNVIDIA	74.5	68.7	81.5	91.4	—	—	32	262K	$0.625	$3.13
31	MiniMax M2.7MiniMax	74.4	71.1	77.7	86.3	78.3	77	29.2	205K	$0.300	$1.20
32	MiMo-V2.5-ProXiaomi	72.8 low data数据少	87.4	—	—	37.3	—	42.2	1.1M	$0.435	$0.870
33	Grok 4.3(high)xAI	72.3	77.7	70.8	—	67.4	98.6	35.8	1M	$1.25	$2.50
34	GLM 5.1(on/off开/关)Z.ai	71.1	66.2	66.9	91.6	68.4	98.6	42	205K	$0.966	$3.04
35	Kimi K2 Thinking(always-on常开)Moonshot AI	70.1	65.5	69.8	92.2	62.5	93.8	32.8	262K	$0.600	$2.50
36	Seed-2.0-LiteByteDance Seed	70.1	66.9	72.9	—	93	95.1	13.2	262K	$0.250	$2.00
37	DeepSeek V3.2(on/off开/关)DeepSeek	69.9	65.3	75.4	89.5	52.2	86.2	31.5	164K	$0.269	$0.400
38	GPT-5.6 Luna Pro(max)OpenAI	69.2 low data数据少	85.4	22	—	—	—	37.8	1.1M	$0.200	$1.20
39	Muse Spark 1.1meta	69.2 low data数据少	86.1	—	—	41	74.8	47.5	1M	$1.25	$4.25
40	ERNIE 4.5 VL 424B A47B Baidu	67 low data数据少	68	49.9	—	78	—	—	123K	$0.420	$1.25
41	Mistral Medium 3.5Mistral AI	64	42.9	69.4	85.6	86.3	95.1	32	262K	$1.50	$7.50
42	gpt-oss-120bOpenAI	63.3	59	56.3	83.2	93.4	—	16.5	131K	$0.037	$0.170
43	Mercury 2Inception	57.2 low data数据少	53.8	28.8	—	91.1	—	—	128K	$0.250	$0.750
44	Gemini 3.1 Flash Lite(high)Google	55.7	60.5	19	82.1	30	—	33.2	1M	$0.250	$1.50
45	Gemini 3.5 Flash Lite(high)Google	54.7	60.4	57.8	—	36	—	39.8	1M	$0.300	$2.50
46	Sonar ProPerplexity	51.9 low data数据少	44.3	—	79.5	29	—	—	200K	$3.00	$15
47	Claude Haiku 4.5(high)Anthropic	51	55.8	59.1	68.4	21.9	39.9	28.5	200K	$1.00	$5.00
48	Command ACohere	49.2	36.5	31.4	74.9	90	85.8	13.7	256K	$2.50	$10
49	Mistral LargeMistral AI	48.7	26.6	52.8	77	64.4	—	29	128K	$2.00	$6.00
50	Mistral Small 4Mistral AI	48.5	31.4	58.7	—	93	41.6	—	262K	$0.150	$0.600
51	Command R+ (08-2024)Cohere	42.2 low data数据少	33	—	79.7	6.7	—	—	128K	$2.50	$10
52	Llama 4 MaverickMeta	41.1	41.5	39.6	84.7	11	18	29.5	1M	$0.200	$0.696
53	Phi 4Microsoft	37.3 low data数据少	32.8	—	75.2	6.7	—	0	16K	$0.070	$0.140
54	Nova Premier 1.0Amazon	33.4 low data数据少	33.4	40.3	—	17	—	—	1M	$2.50	$13
55	Nova Pro 1.0Amazon	25.1 low data数据少	30	—	—	7	—	—	300K	$0.800	$3.20
56	Llama 4 ScoutMeta	23	33.7	23.5	—	3.6	15.6	22.8	1.3M	$0.100	$0.300
—	GPT-5.6 Terra Pro(max)OpenAI	— low data数据少	72.8	77	—	—	—	41.2	1.1M	$2.00	$12
—	LongCat 2.0Meituan	— low data数据少	81.7	—	—	100	—	—	1M	$0.300	$1.20
—	Hunyuan A13B InstructTencent	— low data数据少	94.7	—	—	76.8	—	—	131K	$0.140	$0.570
—	Solar Pro 3Upstage	— low data数据少	75.8	—	—	84.3	—	—	131K	$0.150	$0.600
—	Palmyra X5Writer	— low data数据少 vendor data厂商数据	49.4	—	—	—	—	—	1M	$0.600	$6.00
—	KAT-Coder-Pro V2.5Kwaipilot	— low data数据少 vendor data厂商数据	—	—	—	72.5	—	—	262K	$0.740	$2.96

56 of 62 models have enough data (≥30% benchmark coverage) for a public score. Prices are API list prices per 1M tokens from OpenRouter. "vendor data" means >40% of available scores are vendor self-reported — treat with care. The Chinese board lists models reachable via CN-direct endpoints, ranked by score.56/62 个模型数据充足(benchmark 覆盖 ≥30%)予以出分。价格为 OpenRouter 每百万 token API 标价。"厂商数据" 表示可用成绩中厂商自报占比超 40%,请谨慎参考。中文榜 = 支持国内直连的模型按分数排序。