FreeLLM 数据口径
数据日期： · 今日发现 5 · 模型目录 292 条 · 当前有效资源 76 条（存档共 81 条，其中 5 条已结束；查看每日更新）
口径：模型目录 = /models/all/ 列出的全部记录；当前有效资源 = 未结束的优惠记录（与首页“全部”一致）。全站这些数字由同一份构建数据生成。
为什么创建 FreeLLM?Why we created FreeLLM
我们希望解决的问题The problems we want to solve
信息分散，难以查找Scattered and hard to find
优质的免费 AI 资源散落在各个平台、社区和文档中，用户很难全面、及时地找到适合自己的资源。Quality free AI resources are scattered across platforms, communities and documents.
信息过时，真假难辨Outdated and hard to verify
很多资源已经失效或限制变更，用户需要花费大量时间验证可用性。Resources expire and terms change, so users must spend time checking availability.
降低 AI 使用门槛Lower the barrier to AI
通过系统化的收集、验证和整理，让更多人能够轻松找到并使用优质的 AI 资源。Systematic curation and verification help more people find and use quality AI resources.
我们的收录原则Our inclusion principles
确保每个资源都真实、可用、可靠Every resource should be real, usable and reliable
仅收录官方来源Official sources only
只收录来自官方渠道的免费资源，确保信息的权威性和可靠性。Free resources come from official channels to keep information authoritative and reliable.
区分免费机制Distinguish free access
明确标注免费类型（长期免费、试用、限量免费、学生优惠等），避免误导。We clearly label ongoing free tiers, trials, limited offers and student discounts.
分级证据与真实实测Evidence-backed hands-on tests
官方条件先核验；实际调用过的项目才标记“已实测”，完整评测保留测试日期、环境与原始证据。Official terms are verified first; hands-on and full evaluations are labeled only when they were actually run, with date, environment and evidence.
只发布可复核评分Reproducible evaluations only
没有真实测试就不打分；评分必须绑定测试集、模型 ID、平台、日期和原始结果，不用宣传分冒充实测。No real test means no score. Published scores are tied to a test set, exact model ID, provider, date and raw evidence.
排序方式与权重How ranking works
目录排序使用公开配置的五项评分；权重合计 100%。排序不代表模型质量保证。Directory ordering uses five published factors totaling 100%; a rank is not a quality guarantee.
信息新鲜度 · 35%Freshness · 35%
依据最近核验或目录变更日期，并随时间衰减。Based on the latest verification or catalog change, with a time decay.
免费价值 · 30%Free value · 30%
由免费可用性（40%）、额度（30%）、接入便利度（20%）和核验时效（10%）组成。Combines availability (40%), quota (30%), access convenience (20%), and verification recency (10%).
热度参考 · 20%Popularity reference · 20%
优先使用条目单独配置的热度分；否则使用 Provider 参考值，未配置时采用默认分 50。Uses per-offer value first, then provider reference, and the configured default score when neither is available.
用户行为 · 10%User behavior · 10%
按点击和曝光计算；缺少条目级汇总数据时采用中性分 50，低样本会按置信度向全站基准收缩。Calculated from clicks and impressions; missing item-level aggregate data uses the neutral score of 50, while low sample counts are blended toward the site baseline.
特别优惠 · 5%Special offer · 5%
依据配置的学生优惠、开发者额度和创业支持类别。Based on configured student, developer-credit, and startup categories.
人工加分和扣分目前均未配置；没有付费置顶或商业排名。No manual boosts or deductions are configured. There is no paid placement or commercial ranking.
如何保持内容更新?How we keep content current
自动化监控 + 人工验证，确保信息及时准确Automated monitoring and human checks keep information current
自动监控Automated monitoring
每日自动检查官方页面，检测资源的可用性和变更。Official pages are checked daily for availability and changes.
人工验证Human verification
重要变更由人工再次验证，确保信息准确无误。People verify important changes to keep information accurate.
更新发布Publish updates
通过审核后，及时更新到网站。Approved updates are published promptly.
公开日志Public changelog
所有重要变更都会在更新页面公示。Important changes are listed in the public changelog.
独立性与广告说明Independence and advertising
保持内容的客观性和公正性Keeping our content impartial and fair
不接受付费收录No paid listings
所有资源均基于公开信息收录，不接受任何形式的付费收录。All listings are based on public information. We do not accept payment for inclusion.
不接受付费排名No paid or sponsored ranking
目录会按已公开的评分规则排序；不出售位置，也不接受商业加权。The directory uses its disclosed scoring rules; placement is not sold or commercially weighted.
广告与内容分离Ads stay separate
Google AdSense 广告用于支持运营，与目录内容严格分离，也不代表 FreeLLM 的推荐。详见隐私政策。Google AdSense helps fund the site; ads stay separate from content and do not represent endorsements. See the privacy policy.
社区共建与问题反馈Community and feedback
帮助我们做得更好Help us make FreeLLM better
发现问题?Found a problem?
如果您发现某个资源无法访问、信息有误，或有新的优质资源推荐，欢迎通过以下方式联系我们。你的反馈对我们非常重要！If a resource is unavailable, information is wrong, or you have a recommendation, please let us know.
提交反馈联系我们Contact us
有任何建议、合作意向或其他问题，欢迎通过邮件与我们联系。我们会尽快回复您。For suggestions, partnerships or other questions, email us. We will reply as soon as possible.
提交资源或反馈纠错