评分分布

Q: What is the AI model score distribution?

The score distribution shows how all 290+ tracked AI models are spread across the 0-100 SignalScore scale. Most models cluster in the 40-70 range, with a small elite group scoring above 80 and budget/older models falling below 30.

Q: How is the SignalScore calculated?

SignalScore is a composite metric combining six weighted factors: benchmark performance (90%) from MMLU, GPQA, HumanEval, SWE-bench, and 15+ standardized evaluations, with capabilities and context window as tiebreakers (10%). Each factor is normalized to a 0-100 scale before weighting.

Q: What score percentile is considered good for an AI model?

Models scoring above the 75th percentile (typically 65+ SignalScore) are considered strong performers. The top 10% of models score above 78, while the median score across all tracked models sits around 52-55.

300个AI模型综合评分分布的统计分析。探索均值、中位数、百分位数和层级分布，了解AI模型格局。

关键统计

所有300个评分模型的汇总统计。

平均分

67.8

+/- 18.5 标准差

中位分

71.8

评分范围

40-97

第95百分位

92.2

高于中位数

152

共300个模型

评分分布（10分区间）

LMMarketCap.com

评分分布

每个10分区间中的模型数量。

0-10

10-20

20-30

30-40

40-50

50-60

60-70

70-80

80-90

90-100

Elite

Strong

Average

Below Average

Weak

评分层级

按性能层级分组的模型及汇总统计。

层级	范围	数量	占比	平均评分	顶级模型
Elite	90–100	31	10.3%	92.7	Claude Fable 5(97.1)
Strong	70–89	129	43.0%	80.0	GPT-5.6 Luna Pro(89.0)
Average	50–69	58	19.3%	62.0	Qwen3.5-Flash(68.6)
Below Average	30–49	74	24.7%	40.4	Command R (08-2024)(48.7)
Weak	0–29	0	0.0%	-	-

百分位分析

关键百分位数的评分阈值。

百分位	评分	位置
P5	40.0	4097
P10	40.0	4097
P25	51.2	4097
P50	71.8	4097
P75	82.2	4097
P90	90.4	4097
P95	92.2	4097

按平均分排名的服务商

拥有3+模型的服务商，按平均综合评分排名。

提供商	模型	平均评分	最佳	最差	跨度
1Anthropic	28	82.5	97.1	40.0	57.1
2xAI	5	78.9	88.8	40.0	48.8
3Google	27	78.5	92.2	40.0	52.2
4OpenAI	86	77.6	92.7	40.0	52.7
5MiniMax	8	73.5	78.0	69.1	8.9
6Zhipu AI	13	73.1	78.5	62.5	16.0
7thinkingmachines	3	72.8	74.0	71.6	2.4
8DeepSeek	12	66.4	86.7	40.0	46.7
9Alibaba	31	65.0	80.6	40.0	40.6
10Mistral AI	6	62.3	71.7	40.0	31.7

评分集中度

模型在前20%、中间60%和后20%评分中的分布情况。

前20%(评分 >= 86.7)

64 (21.3%)

中间60%(评分 40.0 - 86.7)

169 (56.3%)

后20%(评分 <= 40.0)

67 (22.3%)

方法论

评分的计算方式以及分布所揭示的信息。

评分计算方式

每个模型获得0到100的评分，主要基于基准测试分数（90%），来源包括Arena Elo、MMLU、GPQA、HumanEval、SWE-bench等15+标准化评估。功能和上下文窗口作为辅助排序（10%）。该评分旨在用一个数字衡量模型的科学评估质量。

分布图告诉我们什么

评分分布揭示了AI模型的竞争格局。中位数附近的紧密聚集表明有许多能力相近的模型，而分散的分布则表明层级之间有明显的差异。分布的形状、偏斜度以及均值和中位数之间的差距都能揭示市场是头重脚轻、底部沉重还是均匀分布。

探索更多

通过基准测试、功能对比和完整排行榜继续探索AI模型数据。

全部探索器基准测试排行榜

Frequently Asked Questions

The score distribution shows how all 290+ tracked AI models are spread across the 0-100 SignalScore scale. Most models cluster in the 40-70 range, with a small elite group scoring above 80 and budget/older models falling below 30.

SignalScore is a composite metric combining six weighted factors: benchmark performance (90%) from MMLU, GPQA, HumanEval, SWE-bench, and 15+ standardized evaluations, with capabilities and context window as tiebreakers (10%). Each factor is normalized to a 0-100 scale before weighting.

Models scoring above the 75th percentile (typically 65+ SignalScore) are considered strong performers. The top 10% of models score above 78, while the median score across all tracked models sits around 52-55.