CHATGPT GEO CRAWL REPORT

ChatGPT 国产大模型 GEO 搜索概率分析报告ChatGPT 国产大模型 GEO 搜索概率分析报告

基于重复采样的 ChatGPT 网页端 AI 搜索结果，识别真正的目标人物或品牌/机构，再衡量提及、平均提及、Top 1 / Top 3 / Top 5 概率、情感倾向，以及信源与标题结构。
实体口径：围绕目标实体识别同类型竞品

Based on repeated ChatGPT web AI search samples, this report identifies the real target person, brand, or organization, then measures mentions, average mentions, Top 1 / Top 3 / Top 5 probability, sentiment, citations, and title patterns.

报告概览Report Overview

概述Overview

本报告基于 5 个关键词、15 次计划采样、15 条有效 ChatGPT AI 搜索结果，评估产品「豆包」的提及、推荐排序、竞品对比和信源结构。目标实体 Top 5 概率为 40.0%，Top 3 概率为 26.7%，平均提及 2.87 次，情感倾向为「积极」，本次识别同类型竞品 14 个，引用信源 196 条。

This report is based on 5 keywords, 15 planned samples, and 15 valid ChatGPT AI search answers. It evaluates the target entity "豆包" across mentions, recommendation ranking, same-type competitors, and citation structure. The target Top 5 probability is 40.0%, Top 3 probability is 26.7%, average mentions per valid answer are 2.87, and the dominant sentiment is Positive. The run identified 14 same-type competitors and 196 citations.

目录Contents

报告概览Overview核心结论Key Findings竞品分析Competitor Analysis情感提及Sentiment实体识别Entity Audit采集覆盖Coverage概率排名Probability信源结构Sources标题特征Titles总结建议Recommendations

指标说明Metric Notes

提及率实体在有效回答中被提到的比例。
平均提及次数实体在每条有效回答中被提到的平均次数。
Top 1 概率实体在推荐排序中位列第 1 的比例。
Top 3 概率实体进入前 3 个推荐位置的比例。
Top 5 概率实体进入前 5 个推荐位置的比例。
平均排名实体出现时的平均推荐位置，数值越低越好。
情感倾向基于实体附近文本词表的启发式判断，分为积极、中性、负向。
信源引用ChatGPT 回答引用的来源标题、域名、渠道和 URL。
Mention RateShare of valid answers where the entity appears.
Average MentionsAverage number of entity mentions per valid answer.
Top 1 ProbabilityShare of valid answers where the entity ranks first.
Top 3 ProbabilityShare of valid answers where the entity enters the top three positions.
Top 5 ProbabilityShare of valid answers where the entity enters the top five positions.
Average RankAverage recommendation rank when the entity appears. Lower is better.
SentimentHeuristic sentiment around the entity mention: positive, neutral, or negative.
CitationSource title, domain, channel, and URL cited by ChatGPT.

目标实体指标数据Target Entity Metrics

提及率73.3%
平均提及2.87
Top 10.0%
Top 326.7%
Top 540.0%
平均排名4.9
情感倾向积极
Mention73.3%
Avg Mentions2.87
Top 10.0%
Top 326.7%
Top 540.0%
Avg Rank4.9
SentimentPositive
5
问题数
chatgpt-crawl.json
15
计划样本
完成 15，未完成 0
15
有效样本
失败 0，有效率 100.0%
196
信源引用
79 个域名 · 105 个 URL
50362
回答字数
完成率 100.0%
5
Questions
chatgpt-crawl.json
15
Planned Samples
completed 15, pending 0
15
Valid Samples
failed 0, valid rate 100.0%
196
Citations
79 domains · 105 URLs
50362
Answer Chars
completion 100.0%

核心结论Key Findings

基于目标实体、同类型竞品、有效样本和引用信源生成最多 10 条高优先级结论，优先呈现目标实体相对竞品的排序与概率差距。

Up to 10 priority findings are generated from the target entity, same-type competitors, valid samples, and cited sources, with emphasis on relative ranking and probability gaps.

目标实体「豆包」提及率 73.3%，Top 5 概率 40.0%，Top 3 概率 26.7%。
目标实体平均每条有效回答提及 2.87 次，情感以「积极」为主。
Top 3 最强竞品是 通义千问，目标实体相对差距 -66.7pp。
竞品识别口径为「产品」，共识别 14 个同类型竞品实体。
Top 1 概率最高的是 通义千问，为 46.7%。
Top 3 覆盖最稳定的是 通义千问，为 93.3%，提及率 100.0%。
Top 5 覆盖最高的是 通义千问，为 100.0%。
信源渠道以 other 为主，占 56.6%。
最常被引用的域名是 jishuzhan.net，出现 8 次。
标题特征中「含数字」最突出，占 80.6%。
Target "豆包" has a mention rate of 73.3%, Top 5 probability of 40.0%, and Top 3 probability of 26.7%.
The target is mentioned 2.87 times per valid answer on average, with dominant sentiment: Positive.
The strongest Top 3 competitor is 通义千问; target gap is -66.7pp.
The competitor scope identified 14 same-type entities.
The leading citation channel is other, accounting for 56.6%.
The most cited domain is 技术栈, with 8 citations.

竞品分析Competitor Analysis

本模块只比较与目标实体类型一致的实体，目标实体「豆包」作为基准，最多展示 1 个目标实体和 9 个同类型竞品。

This section compares only same-type entities. The target entity "豆包" is the baseline, with up to 9 competitors shown.

目标与最佳 3 个竞品 100 分制雷达Target vs Best 3 Competitors Radar

255075100提及提及强度Top 1Top 3Top 5信源行业最优 = 100 分
豆包通义千问DeepSeek文心一言
255075100MentionIntensityTop 1Top 3Top 5CitationsBest in set = 100
豆包通义千问DeepSeek文心一言

Top 5 概率 x 提及率气泡对标Top 5 Probability x Mention Rate Bubble Benchmark

252550507575100100Top 5 概率提及率高覆盖 / 高提及通义千问，Top 5 100.0%，提及率 100.0%，平均提及 10.27通义千问DeepSeek，Top 5 93.3%，提及率 100.0%，平均提及 8.67DeepSeekKimi，Top 5 80.0%，提及率 93.3%，平均提及 4.20Kimi文心一言，Top 5 66.7%，提及率 93.3%，平均提及 2.33文心一言GLM，Top 5 60.0%，提及率 100.0%，平均提及 6.80GLM豆包，Top 5 40.0%，提及率 73.3%，平均提及 2.87豆包混元，Top 5 33.3%，提及率 66.7%，平均提及 2.13混元盘古大模型，Top 5 6.7%，提及率 20.0%，平均提及 0.53盘古大模型阶跃星辰，Top 5 6.7%，提及率 6.7%，平均提及 0.07阶跃星辰MiniMax，Top 5 0.0%，提及率 60.0%，平均提及 1.60MiniMax
通义千问DeepSeekKimi文心一言GLM豆包混元盘古大模型阶跃星辰MiniMax
252550507575100100Top 5 ProbabilityMention RateHigh coverage / high mention通义千问, Top 5 100.0%, mention rate 100.0%, average mentions 10.27通义千问DeepSeek, Top 5 93.3%, mention rate 100.0%, average mentions 8.67DeepSeekKimi, Top 5 80.0%, mention rate 93.3%, average mentions 4.20Kimi文心一言, Top 5 66.7%, mention rate 93.3%, average mentions 2.33文心一言GLM, Top 5 60.0%, mention rate 100.0%, average mentions 6.80GLM豆包, Top 5 40.0%, mention rate 73.3%, average mentions 2.87豆包混元, Top 5 33.3%, mention rate 66.7%, average mentions 2.13混元盘古大模型, Top 5 6.7%, mention rate 20.0%, average mentions 0.53盘古大模型阶跃星辰, Top 5 6.7%, mention rate 6.7%, average mentions 0.07阶跃星辰MiniMax, Top 5 0.0%, mention rate 60.0%, average mentions 1.60MiniMax
通义千问DeepSeekKimi文心一言GLM豆包混元盘古大模型阶跃星辰MiniMax

同类型实体多指标矩阵Same-Type Entity Metric Matrix

实体提及均次Top1Top3Top5
通义千问100.0%10.2746.7%93.3%100.0%
DeepSeek100.0%8.6746.7%60.0%93.3%
Kimi93.3%4.200.0%26.7%80.0%
文心一言93.3%2.336.7%46.7%66.7%
GLM100.0%6.800.0%33.3%60.0%
豆包73.3%2.870.0%26.7%40.0%
混元66.7%2.130.0%13.3%33.3%
盘古大模型20.0%0.530.0%0.0%6.7%
阶跃星辰6.7%0.070.0%0.0%6.7%
MiniMax60.0%1.600.0%0.0%0.0%
EntityMentionAvgTop1Top3Top5
通义千问100.0%10.2746.7%93.3%100.0%
DeepSeek100.0%8.6746.7%60.0%93.3%
Kimi93.3%4.200.0%26.7%80.0%
文心一言93.3%2.336.7%46.7%66.7%
GLM100.0%6.800.0%33.3%60.0%
豆包73.3%2.870.0%26.7%40.0%
混元66.7%2.130.0%13.3%33.3%
盘古大模型20.0%0.530.0%0.0%6.7%
阶跃星辰6.7%0.070.0%0.0%6.7%
MiniMax60.0%1.600.0%0.0%0.0%
角色Role实体Entity提及率Mention Rate平均提及Avg MentionsTop 1Top 3Top 5平均排名Avg Rank信源命中Citation HitsTop 3 差距Top 3 GapTop 5 差距Top 5 Gap
目标Target豆包73.3%2.870.0%26.7%40.0%4.927基准Baseline基准Baseline
竞品Competitor通义千问100.0%10.2746.7%93.3%100.0%1.765+66.7pp+60.0pp
竞品CompetitorDeepSeek100.0%8.6746.7%60.0%93.3%2.764+33.3pp+53.3pp
竞品Competitor文心一言93.3%2.336.7%46.7%66.7%3.819+20.0pp+26.7pp
竞品CompetitorGLM100.0%6.800.0%33.3%60.0%4.742+6.7pp+20.0pp
竞品CompetitorKimi93.3%4.200.0%26.7%80.0%4.4580.0pp+40.0pp
竞品Competitor混元66.7%2.130.0%13.3%33.3%5.52-13.3pp-6.7pp
竞品CompetitorMiniMax60.0%1.600.0%0.0%0.0%7.613-26.7pp-40.0pp
竞品Competitor讯飞星火20.0%0.330.0%0.0%0.0%9.00-26.7pp-40.0pp
竞品Competitor盘古大模型20.0%0.530.0%0.0%6.7%6.70-26.7pp-33.3pp

情感与提及强度Sentiment and Mention Strength

情感分析基于目标实体附近文本的启发式词表判断，平均提及次数统计每条有效回答里同一实体被重复提到的强度；图表默认展示前 10 个实体。

Sentiment is inferred from text near entity mentions, while average mentions measure how often the same entity appears in each valid answer. Charts show up to 10 entities.

目标实体情感分布Target Sentiment Mix

积极100.0%
积极100.0%11
中性0.0%0
负向0.0%0
Positive100.0%
Positive100.0%11
Neutral0.0%0
Negative0.0%0

平均提及次数Average Mentions

通义千问
10.27
DeepSeek
8.67
GLM
6.80
Kimi
4.20
豆包
2.87
文心一言
2.33
混元
2.13
MiniMax
1.60
盘古大模型
0.53
百川智能
0.40

负向占比Negative Share

文心一言
7.1% · 积极
通义千问
0.0% · 积极
DeepSeek
0.0% · 积极
GLM
0.0% · 积极
Kimi
0.0% · 积极
豆包
0.0% · 积极
混元
0.0% · 积极
MiniMax
0.0% · 积极
讯飞星火
0.0% · 积极
盘古大模型
0.0% · 积极
文心一言
7.1% · Positive
通义千问
0.0% · Positive
DeepSeek
0.0% · Positive
GLM
0.0% · Positive
Kimi
0.0% · Positive
豆包
0.0% · Positive
混元
0.0% · Positive
MiniMax
0.0% · Positive
讯飞星火
0.0% · Positive
盘古大模型
0.0% · Positive

目标实体识别Entity Identification

目标类型：产品。语义复核状态：规则回退。系统会把人、公司、产品、概念词和噪声词分开；只有同类型候选项会进入目标与竞品概率计算。自动识别是启发式判断，正式报告建议提供实体别名表并开启 required 语义复核。

Target type: Product. Semantic review status: Rule fallback. The system separates people, companies, products, concepts, and noise terms; only same-type candidates enter target and competitor probability calculations. Automatic recognition is heuristic; formal reports should provide an entity alias list and use required semantic review.

候选项Candidate规则类型Rule Type语义标签Semantic Label同类型Same Type进入竞品In Matrix语义置信度Semantic Confidence样本数Samples提及率Mention RateTop 3Top 5判定理由Decision Reason
通义千问product直接竞品Direct Competitor是Yes是Yes95.0%15100.0%93.3%100.0%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.
DeepSeekproduct待确认Needs Review是Yes是Yes50.0%15100.0%60.0%93.3%候选项证据不足，建议人工确认。The candidate is a same-type entity with answer-body evidence.
文心一言product直接竞品Direct Competitor是Yes是Yes95.0%1493.3%46.7%66.7%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.
GLMproduct直接竞品Direct Competitor是Yes是Yes95.0%15100.0%33.3%60.0%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.
Kimiproduct待确认Needs Review是Yes是Yes50.0%1493.3%26.7%80.0%候选项证据不足，建议人工确认。The candidate is a same-type entity with answer-body evidence.
豆包product目标别名Target Alias是Yes否No96.0%1173.3%26.7%40.0%候选项与目标实体或目标别名匹配。The candidate matches the target entity or one of its aliases.
混元product直接竞品Direct Competitor是Yes是Yes95.0%1066.7%13.3%33.3%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.
盘古大模型product直接竞品Direct Competitor是Yes是Yes95.0%320.0%0.0%6.7%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.
阶跃星辰product直接竞品Direct Competitor是Yes是Yes95.0%16.7%0.0%6.7%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.
MiniMaxproduct直接竞品Direct Competitor是Yes是Yes95.0%960.0%0.0%0.0%候选项与目标实体类型一致，且有正文证据、实体形态和重复样本支撑。The candidate is a same-type entity with answer-body evidence.

采集覆盖Collection Coverage

本次计划采样 15 次，完成 15 次，有效 15 次；下方最多展示 10 个关键词的采样覆盖情况。

This run planned 15 samples, completed 15, and produced 15 valid samples. Up to 10 keyword coverage rows are shown below.

每个问题的有效采样率Valid Sample Rate by Question

q01
3/3
q02
3/3
q03
3/3
q04
3/3
q05
3/3

采样状态构成

有效 100.0%失败 0.0%未完成 0.0%
完成率100.0%
有效率100.0%
平均信源13.1
平均字数3,357

Sample Status Mix

Valid 100.0%Failed 0.0%Pending 0.0%
Completion100.0%
Valid Rate100.0%
Avg Citations13.1
Avg Chars3,357
问题Question问句Prompt完成/计划Done/Planned有效Valid失败Failed未完成Pending平均信源Avg Citations平均回答字数Avg Answer Chars
q01国内目前最值得关注的大模型有哪些？3/330014.03,407
q02中国大模型排名前十的公司和产品分别是什么？3/330012.73,432
q03现在国产大模型里，哪些模型的推理和代码能力最强？3/330013.73,529
q04国产大模型主要可以分为哪些类型，各自适合什么场景？3/330012.03,022
q05如果企业要接入国产大模型，应该优先选择哪些模型？3/330013.03,397

目标实体概率Entity Probability

概率指标来自有效样本中的重复推荐位置估计，图表最多展示 10 个实体；Top 1、Top 3、Top 5 分别观察首位推荐、核心推荐区和扩展推荐区的稳定性。

Probability metrics are estimated from repeated valid samples. Charts show up to 10 entities across Top 1, Top 3, Top 5, mentions, and average rank.

Top 1 概率Top 1 Probability

通义千问
46.7% · 7
DeepSeek
46.7% · 7
文心一言
6.7% · 1
GLM
0.0% · 0
Kimi
0.0% · 0
豆包
0.0% · 0
混元
0.0% · 0
MiniMax
0.0% · 0
讯飞星火
0.0% · 0
盘古大模型
0.0% · 0

Top 3 概率Top 3 Probability

通义千问
93.3% · 14
DeepSeek
60.0% · 9
文心一言
46.7% · 7
GLM
33.3% · 5
Kimi
26.7% · 4
豆包
26.7% · 4
混元
13.3% · 2
MiniMax
0.0% · 0
讯飞星火
0.0% · 0
盘古大模型
0.0% · 0

Top 5 概率Top 5 Probability

通义千问
100.0% · 15
DeepSeek
93.3% · 14
Kimi
80.0% · 12
文心一言
66.7% · 10
GLM
60.0% · 9
豆包
40.0% · 6
混元
33.3% · 5
盘古大模型
6.7% · 1
阶跃星辰
6.7% · 1
MiniMax
0.0% · 0

提及率Mention Rate

通义千问
100.0% · 15
DeepSeek
100.0% · 15
GLM
100.0% · 15
文心一言
93.3% · 14
Kimi
93.3% · 14
豆包
73.3% · 11
混元
66.7% · 10
MiniMax
60.0% · 9
讯飞星火
20.0% · 3
盘古大模型
20.0% · 3

平均提及次数Average Mentions

通义千问
10.27
DeepSeek
8.67
GLM
6.80
Kimi
4.20
豆包
2.87
文心一言
2.33
混元
2.13
MiniMax
1.60
盘古大模型
0.53
百川智能
0.40

平均排名 · 数值越低越好Average Rank · Lower Is Better

通义千问
1.7
DeepSeek
2.7
文心一言
3.8
阶跃星辰
4.0
Kimi
4.4
GLM
4.7
豆包
4.9
混元
5.5
盘古大模型
6.7
MiniMax
7.6

目标实体明细Target Entity Details

实体Entity提及率Mention Rate平均提及Avg MentionsTop 1Top 3Top 5平均排名Avg Rank别名Aliases
通义千问100.0%10.2746.7%93.3%100.0%1.7通义千问 / Qwen / 千问 / Qwen3 / Qwen2.5
DeepSeek100.0%8.6746.7%60.0%93.3%2.7DeepSeek / DeepSeek-V3 / DeepSeek-R1 / 深度求索
文心一言93.3%2.336.7%46.7%66.7%3.8文心一言 / ERNIE Bot / 文心大模型 / ERNIE / ERNIE 4.5
GLM100.0%6.800.0%33.3%60.0%4.7GLM / 智谱GLM / ChatGLM / GLM-4 / GLM-4.5 / 智谱清言
Kimi93.3%4.200.0%26.7%80.0%4.4Kimi / Kimi K2 / Kimi大模型 / Moonshot Kimi
豆包73.3%2.870.0%26.7%40.0%4.9豆包 / Doubao / 豆包大模型 / 豆包模型
混元66.7%2.130.0%13.3%33.3%5.5混元 / 腾讯混元 / Hunyuan
MiniMax60.0%1.600.0%0.0%0.0%7.6MiniMax / abab / 海螺AI / 海螺大模型
讯飞星火20.0%0.330.0%0.0%0.0%9.0讯飞星火 / 星火大模型 / SparkDesk / iFlytek Spark
盘古大模型20.0%0.530.0%0.0%6.7%6.7盘古大模型 / 盘古 / Pangu

问题 x 目标实体 Top 3 热力图Question x Target Entity Top 3 Heatmap

Query通义千问DeepSeek文心一言GLMKimi豆包混元MiniMax讯飞星火盘古大模型Question
q01100.0%33.3%66.7%0.0%33.3%33.3%33.3%0.0%0.0%0.0%国内目前最值得关注的大模型有哪些？
q02100.0%33.3%66.7%0.0%0.0%66.7%33.3%0.0%0.0%0.0%中国大模型排名前十的公司和产品分别是什么？
q03100.0%100.0%0.0%33.3%66.7%0.0%0.0%0.0%0.0%0.0%现在国产大模型里，哪些模型的推理和代码能力最强？
q04100.0%33.3%100.0%66.7%0.0%0.0%0.0%0.0%0.0%0.0%国产大模型主要可以分为哪些类型，各自适合什么场景？
q0566.7%100.0%0.0%66.7%33.3%33.3%0.0%0.0%0.0%0.0%如果企业要接入国产大模型，应该优先选择哪些模型？

信源结构Citation Sources

信源结构用于判断 ChatGPT 回答更依赖官方、媒体、社区、开发者或其他渠道；域名和来源明细默认只展示前 10 条。

Source structure shows whether ChatGPT relies more on official, media, community, developer, or other channels. Domain and source details show up to 10 rows.

高频来源名Frequent Source Names

csdn.net
11
sina.com.cn
10
Reuters
6
Presenc AI
6
ai-learn.cn
6
zishuzhan.com
5
codingplan.fyi
5
jishuzhan.net
4
checkaimodels.com
4
sina.cn
4

高频域名Frequent Domains

技术栈jishuzhan.net
8
Presenc AIpresenc.ai
7
Reutersreuters.com
6
sina.com.cnfinance.sina.com.cn
6
Check.AIcheckaimodels.com
6
智数站zishuzhan.com
6
sina.cnk.sina.cn
6
ai-learn.cnai-learn.cn
6
sina.com.cnk.sina.com.cn
5
codingplan.fyicodingplan.fyi
5

域名占比树图Domain Share Treemap

技术栈8
jishuzhan.net
Presenc AI7
presenc.ai
Reuters6
reuters.com
sina.com.cn6
finance.sina.com.cn
Check.AI6
checkaimodels.com
智数站6
zishuzhan.com
sina.cn6
k.sina.cn
ai-learn.cn6
ai-learn.cn
sina.com.cn5
k.sina.com.cn
codingplan.fyi5
codingplan.fyi

高频 URLFrequent URLs

jishuzhan.net/article/2065289201393954817
8
zishuzhan.com/article-llm-comparison.html
6
ai-learn.cn
6
checkaimodels.com/zh/articles/china-ai-models-landscape-2026
5
bingotech.net/news/2026-Token-DeepSeek-GLM-Kimi.html
4
aigjw.com.cn/news/2026/176874921620.html
4
k.sina.cn/article_7879848900_1d5acf3c401902xa4a.html
4
codingplan.fyi/articles/model_comparisons/20260604
4
xmsumi.com/detail/2984
4
benchlm.ai/blog/posts/best-chinese-llm
4

标题特征Title Patterns

标题特征用于观察被引用内容的表达方式，包括数字、年份、疑问句、结构符和是否包含目标或竞品实体。高频标题最多展示 10 条。

Title patterns show how cited content is framed, including numbers, years, questions, structural punctuation, and entity mentions. Up to 10 frequent titles are shown.

标题功能特征Title Feature Signals

标题含数字
80.6% · 158
标题含年份
66.3% · 130
标题含疑问
19.9% · 39
标题含结构符
49.0% · 96
标题含实体
45.9% · 90
Has Number
80.6% · 158
Has Year
66.3% · 130
Question Form
19.9% · 39
Structured Punctuation
49.0% · 96
Contains Entity
45.9% · 90

标题长度Title Length

25-40
19.9% · 39
41+
73.5% · 144
13-24
5.6% · 11
0-12
1.0% · 2

时间新旧Recency

本年
86.7% · 170
未识别
7.7% · 15
近一年
4.6% · 9
更早
1.0% · 2
Current year
86.7% · 170
Unknown
7.7% · 15
Last year
4.6% · 9
Older
1.0% · 2

标题意图特征分析Title Intent Analysis

趋势新闻
69.4% · 136
对比评测
38.3% · 75
榜单排名
20.4% · 40
其他信息
15.8% · 31
指南攻略
8.7% · 17
推荐选择
3.1% · 6
研究论文
2.0% · 4
避坑风险
1.5% · 3
Trend/News
69.4% · 136
Comparison/Review
38.3% · 75
Ranking/List
20.4% · 40
Other
15.8% · 31
Guide/How-to
8.7% · 17
Recommendation/Choice
3.1% · 6
Research/Paper
2.0% · 4
Risk Avoidance
1.5% · 3

高频引用标题Frequent Cited Titles

#标题Title来源Source次数Count
12026 国产 AI 大模型横评：DeepSeek、通义千问、Kimi、文心一言、星火、豆包谁更能打？ - 技术栈jishuzhan.net8
22026年国内大模型终极横评 - 智数站zishuzhan.com6
3AI Learn | 2026 国产前沿模型观察站 (Frontier AI Radar)ai-learn.cn6
42026 年中国 AI 大模型厂商行业报告|市场份额|文心|应用场景|阿里巴巴|百度_新浪新闻sina.cn5
52026 国产 AI 模型哪个最强？DeepSeek/Qwen/Kimi/GLM/MiniMax 价格+跑分对比 | Check.AIcheckaimodels.com5
62026国产大模型Token成本对比：DeepSeek/GLM/Kimi/通义千问谁性价比最高？bingotech.net4
7国产AI主流模型比较：DeepSeek、kimi、千问、豆包、元宝详细对比与排行榜-行业资讯-AI工具网aigjw.com.cn4
82026年6月主流大模型Coding能力深度对比｜GPT-5.5、Claude Opus 4.8、国产多款跻身前十codingplan.fyi4
9国产大模型横向对比：Kimi K2.6、GLM-5.1、Qwen3、MiniMax M2 四大模型选型指南 - 苏米客苏米客4
10Best Chinese LLMs in 2026: DeepSeek V4, Kimi K2.6, GLM-5, Qwen, and Every Model Ranked | BenchLM.aibenchlm.ai4

总结建议Recommendations

总结建议与 GEO 优化措施

优化优先级

推荐覆盖
100
官方信源
99
标题意图匹配
81
复测归因
55
第三方背书
46
提及强度
35

核心指标趋势预估

0255075100当前优化后1个月3个月6个月
提及率Top 3Top 5

基于当前缺口和优化优先级的保守预估，非真实历史数据。

具体方法与验收指标

维度优先级当前指标具体方法
推荐覆盖100目标 Top 3 26.7%；竞品差距 +66.7pp建设包含「豆包」的对比页、推荐页和场景页，并在标题、首段、对比表和 FAQ 中稳定出现目标实体。
官方信源99官方信源占比 0.5%补充官网说明页、服务页、案例页、流程/价格页和 FAQ，保持 URL 稳定、摘要清晰、实体名称一致。
标题意图匹配81最强标题信号 80.6%优先使用年份、榜单、排名、对比、推荐、疑问和避坑类标题，贴近当前被 ChatGPT 引用的标题表达。
复测归因55跟踪 5 个关键词每周复测同一批关键词，记录内容更新后 Top 3、Top 5、提及强度和引用域名的变化，定位有效页面。
第三方背书46媒体信源占比 16.3%布局行业门户、媒体测评、合作案例、榜单/排名页，统一目标实体名称、别名和引用口径。
提及强度35平均提及 2.87 次让「豆包」进入首段、摘要条目、对比表、结论、图片替代文本和可被引用的短句片段。
建设包含「豆包」的对比页、推荐页和场景页，并在标题、首段、对比表和 FAQ 中稳定出现目标实体。
补充官网说明页、服务页、案例页、流程/价格页和 FAQ，保持 URL 稳定、摘要清晰、实体名称一致。
优先使用年份、榜单、排名、对比、推荐、疑问和避坑类标题，贴近当前被 ChatGPT 引用的标题表达。
每周复测同一批关键词，记录内容更新后 Top 3、Top 5、提及强度和引用域名的变化，定位有效页面。
布局行业门户、媒体测评、合作案例、榜单/排名页，统一目标实体名称、别名和引用口径。
让「豆包」进入首段、摘要条目、对比表、结论、图片替代文本和可被引用的短句片段。

Summary and GEO Optimization Actions

Optimization Priority

Ranking coverage
100
Official citability
99
Title intent fit
81
Measurement loop
55
Third-party proof
46
Mention intensity
35

Core Metric Trend Projection

0255075100NowAfter1M3M6M
MentionTop 3Top 5

Conservative projection from current gaps and action priority; not historical data.

Action Methods and Checks

DimensionPriorityCurrent MetricAction Method
Ranking coverage100Target Top 3 26.7%; competitor gap +66.7ppBuild comparison, recommendation, and use-case pages that explicitly name 豆包 in the title, lead, comparison table, and FAQ.
Official citability99Official source share 0.5%Create official explainer, service, case, pricing/process, and FAQ pages with stable URLs, concise summaries, and clear entity names.
Title intent fit81Strongest title signal 80.6%Prioritize titles with year, ranking, comparison, recommendation, question, and risk-avoidance wording that matches cited-title patterns.
Measurement loop555 keywords trackedRepeat the same keyword set weekly after content updates, then compare Top 3, Top 5, mention intensity, and citation-domain movement.
Third-party proof46Media source share 16.3%Seed industry portals, evaluation articles, partner cases, and list/ranking pages using consistent entity aliases and citations.
Mention intensity35Avg mentions 2.87Place 豆包 in the first paragraph, summary bullets, comparison tables, conclusion, alt text, and reusable source snippets.
Build comparison, recommendation, and use-case pages that explicitly name 豆包 in the title, lead, comparison table, and FAQ.
Create official explainer, service, case, pricing/process, and FAQ pages with stable URLs, concise summaries, and clear entity names.
Prioritize titles with year, ranking, comparison, recommendation, question, and risk-avoidance wording that matches cited-title patterns.
Repeat the same keyword set weekly after content updates, then compare Top 3, Top 5, mention intensity, and citation-domain movement.
Seed industry portals, evaluation articles, partner cases, and list/ranking pages using consistent entity aliases and citations.
Place 豆包 in the first paragraph, summary bullets, comparison tables, conclusion, alt text, and reusable source snippets.