Free Business News API
Selection & Integration Guide免费商业新闻 API
选择与接入指南
Find a free business news API that matches your sources, filters, freshness, usage rights, and production limits—then test it before you build.
从数据源、筛选能力、更新速度、使用许可与免费额度出发，
选择并验证适合项目的免费财经新闻 API。
TL;DR摘要
A usable free tier, relevant business sources, predictable JSON, practical filters, and terms that allow the intended project.
Free may mean no key, a trial, or a recurring quota. Check request limits, delay, history, attribution, and commercial-use rules.
Run the same company, topic, language, and date queries against candidates; inspect relevance, duplicates, timestamps, and missing fields.
QVeris helps agents discover and call external capabilities; each provider’s current data rights and limits still apply.
可持续的免费额度、相关财经来源、稳定 JSON、实用筛选条件，以及允许目标项目使用的条款。
可能是免 Key、限时试用或周期性额度；要核对请求次数、延迟、历史范围、署名与商用限制。
用相同公司、主题、语言和日期查询测试候选接口，检查相关性、重复新闻、时间戳与缺失字段。
QVeris 帮助 Agent 发现并调用外部能力；具体新闻源的实时条款与数据权利仍以供应商为准。
Define the free business news API requirements first先定义免费财经新闻 API 的真实需求
Start by defining whether the product needs a public headline feed, company monitoring, market and industry coverage, economic-policy reporting, or evidence for an agent. Then confirm the access model, returned fields, source and language coverage, freshness, and usage rights. A broad “business” category is useful only when it matches the product's actual decision window and display requirements.
先明确产品需要公开新闻流、公司监控、市场与行业报道、经济政策信息，还是供 Agent 引用的证据；再核对访问方式、返回字段、来源与语言覆盖、时效和使用权。只有当宽泛的“商业新闻”分类真正符合产品的决策窗口和展示要求时，它才有价值。
Coverage and relevance覆盖范围与相关性
Decide whether “business” means broad corporate headlines, company-specific news, financial markets, industries, economic policy, or a regional feed. A general news category may not support ticker-level monitoring.
先明确需要综合商业新闻、公司新闻、金融市场、行业、宏观政策还是特定地区内容。只有“财经”分类的通用接口，未必支持证券代码级筛选。
Freshness, history, and full text更新速度、历史范围与全文
Record timestamp semantics, update cadence, archive window, pagination rules, and whether the API returns metadata, a summary, or licensed article text. Test “real-time” rather than assuming it.
核对时间戳含义、更新频率、历史窗口、分页规则，以及返回的是元数据、摘要还是获得许可的正文。不要仅凭宣传语认定接口是实时的。
Free tier and usage rights免费额度与使用许可
Check calls per minute and month, allowed environments, caching, attribution, redistribution, commercial use, and upgrade behavior. The free plan must fit both your test volume and legal use case.
核对分钟/月调用量、允许环境、缓存、署名、再分发、商用与升级规则。免费方案既要满足测试流量，也要允许你的具体使用方式。
Article identity, syndication, and corrections文章身份、转载关系与更正
A single announcement may appear under a wire service, several publisher URLs, and rewritten headlines. Prefer a provider article ID and canonical URL, but retain publisher, original source, publication time, update time, and language. If the API exposes corrections, retractions, or updated versions, decide whether your application replaces the old record, links the versions, or alerts a reviewer.
同一则公告可能先由通讯社发布，再被多家媒体转载，并出现不同标题。应优先使用供应商文章 ID 和规范链接，同时保留发布媒体、原始来源、首次发布时间、更新时间与语言。若接口提供更正、撤稿或修订版本，还要提前决定是覆盖旧记录、关联版本，还是交给人工复核。
Primary evidence versus reporting一手材料与媒体报道
Press releases, regulatory filings, exchange notices, central-bank statements, and company investor-relations posts are primary evidence; publisher articles interpret or contextualize them. A useful feed identifies the source type and preserves a link to the original material when available. For research or agent-generated answers, keep the reporting article and the primary document as separate records so citations do not imply that commentary is an official disclosure.
新闻稿、监管文件、交易所公告、央行声明和公司投资者关系页面属于一手材料，媒体文章则提供解读与背景。实用接口应标明来源类型，并在可能时保留原始材料链接。用于研究或 Agent 回答时，媒体报道和原始文件应作为两条不同记录保存，避免引用时把评论性内容误写成官方披露。
Entity, event, and language semantics实体、事件与语言语义
Company names, tickers, products, executives, industries, countries, and event types should be distinct fields rather than one keyword list. Verify how aliases, renamed companies, dual listings, ambiguous tickers, and translated names are resolved. If the provider translates headlines or summaries, retain the original language and text, translation status, and model or method; otherwise an English query may hide that the underlying evidence was published in another language.
公司名称、股票代码、产品、高管、行业、国家和事件类型应是不同字段，而不是混在一组关键词里。需要验证公司别名、更名、双重上市、歧义代码和译名如何解析。若供应商翻译标题或摘要，应保留原文语言、原始文本、翻译状态以及所用模型或方法，否则英文查询结果可能掩盖证据最初是以另一种语言发布的。
Match the feed to the product surface让新闻能力匹配产品场景
A public headline widget needs display rights, attribution, images, stable URLs, and aggressive duplicate control. A company-monitoring alert needs low publication-to-API delay, exchange-aware entity matching, event classification, and correction updates. A research assistant needs longer history, primary-source links, full query replay, and enough context to explain why evidence was selected. A model-training or archive workflow adds retention and redistribution requirements. Score providers against the intended surface instead of treating one broad “business news” endpoint as equally suitable for all four.
公开新闻组件需要明确的展示权、署名、图片、稳定链接和严格去重；公司监控提醒更看重从发布到接口可见的延迟、带交易所的实体匹配、事件分类和更正更新；研究助手需要更长历史、一手材料链接、完整查询重放，以及足够解释证据为何被选中的上下文；模型训练或归档还会增加保存与再分发要求。应根据最终产品场景给供应商评分，而不是认为一个宽泛的“商业新闻”端点能同样满足四类需求。
How to compare free business news API options如何比较免费商业新闻 API 选项
| Criterion标准 | Verify核对内容 | Why it matters为什么重要 |
|---|---|---|
| Sources来源 | Publishers, countries, languages, categories媒体、国家、语言、财经分类 | Relevance and geographic blind spots相关性与地域盲区 |
| Filters筛选 | Company, ticker, topic, source, language, date公司、代码、主题、来源、语言、日期 | Less noisy post-processing减少后处理噪声 |
| Response响应 | Stable IDs, URLs, timestamps, summaries, images稳定 ID、链接、时间、摘要、图片 | Deduplication and display去重与展示 |
| Free limits免费限制 | Quota, throttling, delay, history, CORS, key额度、限速、延迟、历史、CORS、Key | Demo versus sustainable integration演示与可持续接入的区别 |
| Rights权利 | Attribution, caching, commercial use, redistribution署名、缓存、商用、再分发 | Technically and legally usable技术与条款都可用 |
Business news APIs to test first可以优先测试的商业新闻 API
The services below represent different retrieval models. Confirm current pricing, free-plan restrictions, source rights, history, and production use on each official site; a development key does not automatically authorize a public or commercial news product.
下面几家代表不同的新闻检索模式。应在各自官方网站核实当前价格、免费层限制、来源授权、历史范围和生产用途；开发测试 Key 并不自动允许制作公开或商业新闻产品。
| Provider供应商 | Good starting point适合从哪里开始 | Best fit更适合 | Verify carefully重点核实 |
|---|---|---|---|
| NewsAPI | Broad publisher discovery through top-headline and article-search endpoints with source, keyword, language, and date-style filters.通过头条与文章搜索端点发现广泛媒体内容，并使用来源、关键词、语言和日期类筛选。 | Evaluating a general business-news feed or quickly testing source and query coverage.评估综合商业新闻流，或快速测试来源与查询覆盖。 | Development versus production rights, article-content fields, result delay, archive depth, source list, attribution, and redistribution.开发与生产权限、文章内容字段、结果延迟、历史深度、来源列表、署名和再分发。 |
| GNews | General news search and top-headline retrieval with language, country, topic, time, and pagination controls.提供综合新闻搜索与头条检索，并支持语言、国家、主题、时间和分页控制。 | Regional or multilingual business-news prototypes that need a simple REST response.需要简单 REST 响应的区域或多语言商业新闻原型。 | Exact business-topic coverage, number of accessible articles, historical limits, content truncation, images, attribution, and commercial display.商业主题覆盖、可访问文章数量、历史限制、正文截断、图片、署名和商业展示权。 |
| Marketaux | Financial news enriched with companies and entities, symbol filters, industries, countries, languages, and sentiment-related fields.提供带公司与实体增强信息的金融新闻，并支持证券代码、行业、国家、语言和情绪字段。 | Ticker-level company monitoring, entity-aware filtering, and financial-news research.证券代码级公司监控、实体筛选和金融新闻研究。 | Entity-link precision, duplicate stories, pagination cost, source mix, historical depth, text storage, and plan-specific limits.实体关联准确率、重复报道、翻页成本、来源构成、历史深度、正文保存权和套餐限制。 |
| Alpha Vantage | A News & Sentiment function focused on market topics and ticker-related articles within a broader financial-data API family.在更广泛的金融数据 API 中提供 News & Sentiment 函数，聚焦市场主题和股票相关文章。 | Scheduled company or market briefs that want news metadata and sentiment alongside other financial endpoints.希望把新闻元数据、情绪和其他金融端点结合起来的定时公司或市场简报。 | Ticker relevance, topic filters, sort order, time range, result limit, source coverage, endpoint access, and whether sentiment is item- or ticker-specific.股票关联度、主题筛选、排序、时间范围、返回数量、来源覆盖、端点权限，以及情绪属于文章级还是股票级。 |
Do not compare only article counts. A smaller feed with stable IDs, relevant company matching, correction delivery, and usable rights can be more valuable than a larger feed dominated by syndicated duplicates or content that cannot be displayed.
不要只比较文章数量。一个新闻流即使规模较小，只要具备稳定 ID、可靠的公司匹配、更正传递和可用授权，也可能比充满转载重复、且无法合法展示的大型新闻流更有价值。
Business news API acceptance examples商业新闻 API 验收实例
One wire story can appear as twenty articles一条通讯社报道可能变成二十篇“新闻”
A company announcement may be distributed by a wire service, republished by regional outlets, copied into finance portals, and summarized under rewritten headlines. Counting every URL as independent evidence inflates source diversity, article volume, and sentiment confidence. Build a duplicate cluster from provider IDs, canonical URLs, original source, normalized headline, publication time, named entities, and text similarity. Keep every permitted record, but choose one representative item and report both raw count and independent-story count.
一则公司公告可能先由通讯社发布，随后被地方媒体转载、进入财经门户，并以改写标题再次摘要。如果把每个 URL 都当作独立证据，来源多样性、文章数量和情绪置信度都会被虚增。应结合供应商 ID、规范 URL、原始来源、标准化标题、发布时间、命名实体和文本相似度建立重复新闻簇。条款允许的记录可以全部保存，但展示时只选一条代表内容，并同时报告原始文章数与独立事件数。
“Apple” is not always Apple Inc.“Apple”并不总是指苹果公司
A keyword query for Apple can retrieve fruit prices, an orchard company, a music label, or an article that mentions an iPhone only in passing. Test exact ticker filters separately from full-text keywords. Require company entity ID, exchange-aware ticker, relevance score, matched text span, and article focus when available. Include negative examples in the benchmark; a feed that returns nothing for an ambiguous query can be safer than one that confidently links the wrong issuer.
只用 Apple 作为关键词，可能检索到水果价格、果园公司、音乐厂牌，或只顺带提到 iPhone 的文章。测试时应把精确证券代码筛选与全文关键词检索分开，并尽量要求返回公司实体 ID、带交易所的证券代码、关联度、命中文本和文章焦点。基准样本必须包含负向案例；对于歧义查询，宁可返回空结果，也比自信地关联到错误发行人更安全。
Prioritize display rights, attribution, stable article URLs, publisher identity, image rights, correction handling, and duplicate control. Do not expose provider-only snippets or images merely because the JSON contains them.
Prioritize publication-to-API latency, entity precision, event classification, source quality, and update or retraction delivery. Define a quiet-period baseline so ordinary mentions do not trigger the same alert as earnings, litigation, or an acquisition.
Require longer history, query replay, primary-source links, article timestamps, source type, duplicate clusters, and enough licensed context to explain selection. Preserve conflicting coverage instead of collapsing it into one summary.
Return article ID, headline, publisher, original source when known, publication and ingestion times, entities, relevance, language, correction state, and citation URL. The agent should separate reported claims from primary evidence and state when coverage is incomplete.
优先检查展示权、署名要求、稳定文章链接、媒体身份、图片授权、更正处理和去重。JSON 中出现摘要或图片，并不意味着可以直接公开展示。
优先检查从发布到 API 可见的延迟、实体匹配准确率、事件分类、来源质量以及更新和撤稿传递。还要建立消息清淡期基线，避免普通提及与财报、诉讼或并购触发同等级告警。
需要更长历史、可重放查询、一手资料链接、文章时间、来源类型、重复新闻簇，以及足以解释筛选原因的合规上下文。相互矛盾的报道应分别保留，不能强行合成一个看似统一的结论。
结果应返回文章 ID、标题、媒体、已知原始来源、发布与收录时间、实体、关联度、语言、更正状态和引用链接。Agent 需要区分媒体报道与一手证据，并在覆盖不完整时明确说明。
Integrate a free business news API step by step免费财经新闻 API 接入步骤
Include a large company, a smaller company, an industry topic, two languages, a date boundary, and an intentionally empty query.
Store credentials server-side, set a timeout, and log the status code, response time, rate-limit headers, and raw response during evaluation.
Map provider fields into your article model. Prefer stable IDs; otherwise combine canonical URL, normalized title, source, and publication time.
Handle 429, malformed items, missing timestamps, empty pages, and outages. Cache only where provider terms allow it.
Store the complete permitted result set before applying your own relevance, freshness, source-quality, or event-impact ranking. Keep the ranking version and feature values so a reviewer can explain why one article appeared above another. Do not let a sentiment score silently become a relevance score.
Replay known queries on a schedule and track publisher coverage, latency percentiles, duplicate groups, retractions, schema changes, and missing fields. Alert when a major source disappears or a cursor starts skipping records; a 200 response can still contain a degraded feed.
包含大型公司、小型公司、行业主题、两种语言、日期边界和一个预期为空的查询。
把凭证保存在服务端，设置超时，并记录状态码、响应时间、限流响应头与原始响应。
把供应商字段映射到自己的新闻模型。优先使用稳定 ID，否则结合规范链接、标准化标题、来源与发布时间。
处理 429、异常条目、缺失时间、空分页与服务中断；只在供应商条款允许时缓存数据。
先保存条款允许范围内的完整结果，再应用自己的相关性、新鲜度、来源质量或事件影响排序。保留排序版本与特征值，让复核人员能够解释某篇文章为何排在另一篇之前。不要把情绪分值静默当作相关性分值。
定期重放已知查询，跟踪媒体覆盖、延迟分位数、重复新闻簇、撤稿、Schema 变化与字段缺失。重要来源消失或游标开始漏记录时应立即告警；HTTP 200 并不代表新闻流质量没有下降。
Use measurable acceptance criteria. For a fixed query set, record relevant articles found, irrelevant results, duplicate-cluster rate, median and 95th-percentile publication-to-API delay, missing-field rate, paging consistency, correction delivery, and source concentration. Include known events that should be found and negative queries that should stay empty. Review false positives by cause—ambiguous entity, stale article, broad category, translation, or syndication—because each needs a different fix. A provider with more headlines is not automatically better if most are duplicates, arrive after the decision window, come from one narrow source group, or cannot be legally displayed. Save the dated scorecard and raw samples so plan or coverage changes can be measured against the original approval evidence.
用可量化标准验收。针对固定查询集，记录相关报道召回、无关结果、重复新闻簇比例、从发布到接口可见延迟的中位数与第 95 百分位、字段缺失率、分页一致性、更正传递情况和来源集中度。样本既要包含必须找到的已知事件，也要包含应保持为空的负向查询。误报应按原因拆分：实体歧义、旧闻、分类过宽、翻译或重复转载，因为每类问题的修复方式不同。若大量结果只是重复内容、错过决策窗口、集中于少数来源，或条款不允许展示，那么“新闻条数更多”并不代表接口更好。保存带日期的评分表与原始样本，后续才能判断套餐或覆盖变化是否偏离最初批准依据。
Use QVeris to discover business news capabilities用 QVeris 发现商业新闻能力
QVeris can help an AI agent discover, inspect, and call relevant external capabilities instead of hard-coding every provider path. It does not change a provider’s coverage, licensing, quota, or freshness.
QVeris 可帮助 AI Agent 发现、检查并调用相关外部能力，减少硬编码供应商路径。它不会改变供应商的覆盖、许可、额度或更新速度。
- Explore the QVeris provider details for news and financial data capabilities.
- Inspect the capability definition and inputs before an agent calls it.
- Keep provider limits, attribution, and validation in application policy.
- 通过 QVeris 工具发现器查找新闻和金融数据能力。
- 让 Agent 调用前先检查能力定义、输入字段与输出结构。
- 在应用策略里保留供应商限制、署名要求与结果验证。
FAQ常见问题
Yes. Some providers offer a free tier or no-key endpoint, but coverage, limits, licensing, article content, and production rights differ.
Some APIs do. Confirm whether matching uses named entities, exact tickers, full-text keywords, or provider tags.
Only when the terms permit your use case and quota, attribution, redistribution, and reliability fit production.
Test representative queries and compare sources, freshness, deduplication, filters, fields, limits, rights, and failures.
有些供应商提供免费层或免 Key 接口，但覆盖、额度、许可、正文字段与生产使用权并不相同。
部分接口支持。要确认它使用实体识别、精确代码、全文关键词还是供应商标签进行匹配。
只有当条款允许你的用途，且额度、署名、再分发与可靠性条件满足生产需求时才可以。
使用代表性查询实测来源、更新速度、去重、筛选、字段、限流、使用权与失败行为。