MODEL BENCHMARKS

模型榜单

查阅国内外主流模型的评测表现，按能力维度比较。

LiveBench 2.0数据版本 2026-10-25

收录模型: 32个
国内模型: 9个
评测维度: 7项

数据核验 2026/11/01 00:52 北京时间评测说明

模型跑分榜

LiveBench 2.0 · 得分范围 0–100，越高越好

查看原榜

32 个模型 · 按综合得分降序筛选保留全榜位次 · 同分并列

LiveBench 2.0 模型排名；缺失分数显示为破折号，不计为零。
位次	模型 / 开发方									覆盖
01	Claude Fable 5.1 Anthropic海外推理强度 max	85.69	77.90	91.70	97.70	89.80	80.30	82.00	83.40	7/7
02	GPT-6 Astra OpenAI海外推理强度 max	85.64	77.50	92.70	93.70	91.30	83.00	85.20	82.00	7/7
03	Claude Opus 5.5 Anthropic海外推理强度 max	85.14	72.80	92.20	96.30	88.40	80.30	85.00	86.90	7/7
04	GPT-6.1 Sol OpenAI海外推理强度 max	84.07	68.70	92.60	96.00	93.30	82.70	83.50	82.10	7/7
05	Claude Sonnet 5.5 Anthropic海外推理强度 max	80.04	63.40	91.60	94.70	89.30	59.50	83.60	85.00	7/7
06	Gemini 4 Argon Google海外推理强度 high	79.30	61.20	95.10	89.70	82.80	85.20	83.10	73.70	7/7
07	GPT-6.1 Terra OpenAI海外推理强度 max	77.79	48.60	88.70	94.30	93.10	82.30	73.90	82.30	7/7
08	Muse Wonder 1.0 Meta海外推理强度 max	74.15	52.60	90.60	86.20	60.00	81.60	88.00	75.10	7/7
09	DeepSeek V4.1 Flash 1017 DeepSeek国内推理强度 max	72.87	35.70	89.10	96.00	67.20	79.50	79.50	83.40	7/7
10	Kimi K3.1 Pro Moonshot AI国内推理强度 max	71.29	39.10	91.10	81.80	75.90	79.10	77.90	77.80	7/7
11	Grok 4.7 xAI海外推理强度 xhigh	70.14	46.10	82.70	84.50	49.30	76.90	86.40	78.00	7/7
12	Claude Haiku 5.5 Anthropic海外推理强度 max	69.85	40.00	82.40	91.40	69.50	53.10	81.90	83.10	7/7
13	MiMo V2.6 Pro Xiaomi国内推理强度 high	68.48	35.20	79.60	81.50	77.30	72.30	76.70	78.50	7/7
14	GPT-5.6 Terra OpenAI海外推理强度 max	67.79	33.10	90.60	86.20	70.30	79.30	60.40	76.00	7/7
15	Grok 4.6 xAI海外推理强度 xhigh	67.75	28.00	91.20	89.10	49.40	75.30	84.50	79.80	7/7
16	MiniMax M3.1 Flash MiniMax国内推理强度 max	66.89	28.80	82.90	83.90	80.60	71.40	72.70	74.00	7/7
17	GPT-6 Luna OpenAI海外推理强度 max	64.59	30.90	81.80	84.50	70.30	73.40	43.50	82.50	7/7
18	Kimi K3 Moonshot AI国内推理强度 max	64.57	26.40	90.70	59.30	75.20	78.70	79.00	77.20	7/7
19	Mistral Large 4 1029 Mistral AI海外推理强度 high	63.49	27.50	85.20	95.40	17.40	76.60	79.70	76.30	7/7
20	Muse Spark 1.3 Meta海外推理强度 max	63.03	25.50	89.70	90.20	30.90	79.60	68.60	74.80	7/7
21	GLM-5.3 Flash Z.AI国内推理强度 max	62.07	19.00	77.60	80.60	59.80	76.40	70.10	77.20	7/7
22	Qwen3.8 Max Alibaba国内推理强度 max	61.56	29.90	88.20	60.30	49.40	78.40	69.90	78.60	7/7
23	Gemini 3.8 Flash Google海外推理强度 high	59.21	16.50	89.30	82.20	45.10	54.00	79.10	72.70	7/7
24	Step 5 (Preview) StepFun国内推理强度 max	58.61	19.40	83.80	85.60	24.80	63.50	72.80	76.90	7/7
25	Qwen3.8 27B Alibaba国内推理强度 max	58.38	20.60	80.00	86.80	24.20	76.60	59.40	75.90	7/7
26	Muse Spark Open Meta海外推理强度 max	51.63	11.70	73.70	73.60	26.20	60.20	80.30	60.30	7/7
27	Smaug-Nemotron-3.5-NVFP4 Abacus.AI海外推理强度 max	50.31	8.40	75.10	72.20	25.60	65.10	67.80	62.70	7/7
28	Nemotron 3.5 Super VL NVIDIA海外推理强度 max	50.23	11.70	74.60	71.30	21.60	64.60	66.70	62.80	7/7
29	inkling 来源未标注待核对推理配置未标注	47.05	8.50	78.30	64.00	12.00	72.80	59.90	58.30	7/7
30	Nemotron 3.5 Super VL (NVFP4) NVIDIA海外推理强度 max	46.50	7.30	71.70	61.30	18.00	65.70	70.90	57.10	7/7
31	Mistral Large 4 Mistral AI海外推理强度 high	43.90	8.80	83.90	60.00	0.60	76.50	28.10	65.60	7/7
32	Gemma 4 31B Google海外推理强度 max	38.72	10.70	68.30	49.00	0.00	53.30	44.40	58.40	7/7