Best Financial News Sentiment APIs in 2026: Ticker-Tagged News Scoring Compared
Ticker-tagged news sentiment ranges from $0 self-serve APIs to six-figure enterprise NLP feeds. We compare eleven developer options on what they score, how far back the history goes, and which ones ship no sentiment field at all.
The short version
- Global multi-asset: Marketaux
- Cheapest one call: Alpha Vantage
- Source-aware global equity/ETF: Adanos Stock News Sentiment API
- Deep data + news: Polygon.io (Massive)
- Bring your own model: Benzinga News API
- Ready-made daily score: StockNewsAPI.com
How they compare
Scan the field, then read the full write-ups below. Pricing and limits were checked in June 2026; vendors change them often, so confirm before you commit.
| Tool | Sentiment granularity | Asset/geographic coverage | Free tier & entry price | Sources & update latency | History depth | Methodology transparency |
|---|---|---|---|---|---|---|
| Marketaux | Per-entity + passage | Global; stk/crypto/fx | Free; $29/mo | 5,000+; poll/REST | Not clearly stated | Proprietary, -1..1 |
| Alpha Vantage News & Sentiment API | Overall + per-ticker | Equities/crypto/fx | 25/day; $49.99/mo | 50+; REST | ~2022+ confirmed | Thresholds public |
| Stockdata.org | Per-entity + highlight | Global claimed; US? | Free; $29/mo | 5,000+; REST | Unclear/disputed | Avg of highlights |
| Adanos Stock News Sentiment API | Per-ticker + raw rows | Global stk/ETF | Free; $29/mo | 50+; 10-min refresh | Up to 365 days | VADER+RoBERTa pub. |
| StockNewsAPI.com | Article + daily agg | US equities | Trial; $19.99/mo | 50+; not stated | Back to 2019 | Keyword/title |
| EOD Historical Data (EODHD) | Article + daily agg | Stk/ETF/fx/crypto | ~2-4 news/day; €19.99/mo | '10+/40+'; 15-60m | Not documented | Lexicon polarity |
| Finnhub, News Sentiment API | Per-company agg | US only | 60/min; paid uncertain | Aggregated; REST | Not verifiable | Undocumented |
| Polygon.io (now Massive) | Per-ticker label | US-centric | Free; $29/mo | Multi-pub; hourly | To 2016 (paid) | AI label, no score |
| Financial Modeling Prep (FMP) | Legacy label only | US (global Ultimate) | No news on free; $19/mo | 3rd-party; hourly | ~2019+ | Opaque/legacy |
| Benzinga News API | None native (DIY) | US/Canada | Free eval; sales | In-house; real-time | Back to 2010 | N/A (no score) |
| Tiingo News API | None (DIY) | US equities | Free; $30/mo | 120+; REST | Back to 2014 | N/A (no score) |
How we ranked them
News sentiment scores are not comparable across vendors, so we ranked on transparency, coverage, and price rather than treating any number as ground truth.
Sources are listed at the foot of this page. Figures reflect June 2026 and vendors change them often.
Article vs entity scoring
We note whether a tool scores the whole article, each tagged ticker, or just a daily aggregate per symbol. That choice shapes what the score can tell you.
Does it score at all
Several popular “news APIs” return tagged news with no sentiment field. We flag those clearly so you do not pay for a number that is not there.
Coverage and history
US-only versus global, equities versus crypto and forex, and how far the archive reaches for backtesting are the sharpest dividing lines, so they carry weight here.
The tools, ranked
The field splits between APIs that return a ready-made score and feeds that hand you clean tagged news to run your own model. Both appear below, labelled, with prices checked in June 2026.
Marketaux
Stock news sentiment APIThe most well-rounded self-serve pick for ready-made news sentiment: genuine per-entity scoring with passage-level highlights across global stocks, crypto and forex at a low price.
- Sources
- 5,000+ global news publishers, Proprietary entity-recognition engine, Proprietary per-entity sentiment scoring
- Access
- REST API (HTTP GET, JSON), API-key auth, Market Stats API for aggregated sentiment
- Best for
- Builders needing broad international, multi-asset news with per-ticker sentiment and passage-level highlights at a low price.
Strengths
- Per-entity sentiment_score (-1 to 1) plus a 'highlights' array showing the exact passages where each ticker is named, each with its own score
- Broad international breadth: 5,000+ sources, 30+ languages, 80+ markets, 200,000+ entities
- Multi-asset in one API (stocks, crypto, forex)
- Usable free tier and low $29/mo entry for testing and small projects
Watch-outs
- Free tier caps each news request to 3 articles, fine for testing, not backfill or dashboards
- REST/polling only, no WebSocket/push, so 'real-time' is poll-based with inherent latency
- Returns summaries/metadata, not full article text
- Sentiment methodology is proprietary with limited public benchmarks
Alpha Vantage News & Sentiment API
Free/cheap sentiment APIThe easiest single-endpoint way to pull news pre-tagged with both an overall and per-ticker sentiment score across equities, crypto and forex, held back only by a very tight free tier.
- Sources
- 50+ major financial news outlets (Benzinga, Motley Fool, Zacks observed)
- Access
- REST/JSON (function=NEWS_SENTIMENT), Official Alpha Vantage MCP server, Wide community SDK support
- Best for
- Indie devs and quants who want a cheap, one-call news feed with overall and per-ticker sentiment without building an NLP pipeline.
Strengths
- Built-in per-entity ticker_sentiment with relevance_score and documented 5-bucket thresholds (Bearish to Bullish)
- Covers equities plus crypto (CRYPTO:) and forex (FOREX:) prefixes in one call
- Flexible filters: tickers, topics, time_from/to, sort, limit up to 1000
- Simple REST/JSON with wide community SDKs and an official MCP server
Watch-outs
- Free tier is only 25 requests/day total, impractical for production news monitoring
- Tiers scale on req/min, so heavy historical backfills are slow/costly
- Relevance scoring criticized as noisy; users report incomplete news lists
- Single proprietary score with coarse labels and no model transparency or low-tier SLA
Stockdata.org
Cheap news + sentiment APIMarketaux-style per-entity sentiment with rich filtering and a free tier, but a thin and partly negative public reputation plus reports of a shrinking offering temper confidence.
- Sources
- 5,000+ news sources in 30+ languages, Market/exchange price data feeds
- Access
- REST API (JSON), Token as GET parameter, Web dashboard
- Best for
- Indie builders who want cheap, ticker-tagged news with per-entity sentiment for a side project or lightweight signal where budget beats deep history.
Strengths
- Per-entity sentiment_score (-1 to +1) plus match_score and per-highlight sentiment with the exact text spans scored
- Rich news filtering: symbols, entity_types, industries, countries, languages, sentiment ranges, domains/source_ids
- Cheap and low-friction free tier (100 req/day) plus stock price data in the same API
- Trending/stats endpoints rank entities by news buzz/sentiment over time windows
Watch-outs
- A Trustpilot reviewer reports the offering shrank (EOD history 40yr to 5, exchanges reduced to US-only) at the same price
- Very thin, negative public reputation (~3.2, from a single 1-star review)
- Tight low-tier article caps (2 free / 10 Basic / 25 Standard per request)
- Historical news depth, latency and current exchange coverage are not clearly documented
A transparent, source-aware global equity/ETF news sentiment feed with a published hybrid model, fast refresh and AI explanations, strong for buzz dashboards, but narrower in asset class and history than the breadth leaders.
- Sources
- 50+ curated financial news feeds canonicalized into source ids (e.g. Reuters, Benzinga, Finviz), Hybrid VADER + Twitter-RoBERTa sentiment model, Groq llama-3.1-8b for trend explanations
- Access
- REST API (JSON/UTF-8), X-API-Key auth, Trending, stock detail, compare (up to 10), market-sentiment endpoints, AI explain endpoint; raw mentions endpoint (Professional)
- Best for
- Builders and retail/quant dashboards wanting source-weighted global stock/ETF news buzz and sentiment with transparent methodology and AI trend explanations.
Strengths
- Transparent published methodology: hybrid VADER + Twitter-RoBERTa per-mention scoring (-1 to +1) with positive/negative/neutral labels, plus a documented 0-100 BuzzScore
- Source-aware by design: distinct publisher counts on trending, detail and market-sentiment endpoints separate one-source noise from broad coverage
- Fast 10-minute refresh and AI trend explanations (Groq llama-3.1-8b) on a clean REST/JSON surface across 35,000+ tickers and 50+ sources
- Professional tier exposes raw article-level mention rows (snippet, source, timestamp) for evidence behind aggregate scores
Watch-outs
- Narrower than breadth leaders: stocks and ETFs only, with no native crypto or forex in this specific product (those live in separate Adanos sentiment APIs)
- Aggregates per ticker rather than offering Marketaux/Stockdata-style multi-entity passage tagging within a single article
- Newer brand with a shorter public track record and retail/social-sentiment roots, not enterprise event-NLP heritage
- Free and Hobby tiers are non-commercial; commercial rights and raw mentions require the $299/mo Professional plan, and history caps at 365 days vs multi-year archives elsewhere
StockNewsAPI.com
Stock news sentiment APICheap, curated US stock news with per-article labels and a daily aggregate sentiment score and 2019 history, but sentiment is keyword/title-based and there is no permanent free tier.
- Sources
- 50+ outlets (CNBC, Bloomberg, Zacks, Motley Fool, Fox Business, TheStreet), Video news; analyst upgrades/downgrades and earnings data
- Access
- REST API (GET, JSON), CSV export via csv parameter
- Best for
- Builders and quant/retail dashboards needing cheap, ticker-tagged US stock news with a ready-made sentiment label and aggregate daily score.
Strengths
- Per-article positive/negative/neutral labels plus a dedicated -1.5 to +1.5 daily Sentiment Score endpoint for tickers and the market
- Curated, recognizable sources (CNBC, Bloomberg, Zacks, Motley Fool) with strict ticker-relevance filtering
- Low, transparent pricing ($19.99/$49.99/mo) with a 5-day full-feature trial
- History back to 2019 plus extras: video, top mentions, upgrades/downgrades, earnings calendar, CSV export
Watch-outs
- Sentiment is keyword/title-based, not ML/NLP, so it is coarse with a large Neutral bucket
- No permanent free tier; trial is only 5 days / 100 calls
- US-equity-focused; crypto/forex need separate sister APIs
- No published per-second rate limits or update-latency SLA
EOD Historical Data (EODHD)
Cheap news + sentiment APITwo complementary outputs, per-article polarity plus a daily aggregated -1 to 1 score across stocks/ETFs/FX/crypto, at a very low price, but with a 15-60 min delay and near-unusable free tier.
- Sources
- Aggregated public financial news portals ('10+ sources with 40+ media'), EODHD sentiment scoring from news (and, per EODHD, social)
- Access
- REST API (JSON, XML option for news), Single key across all EODHD APIs, Python library and Postman collections
- Best for
- Cost-sensitive builders wanting article-level news plus ready-made daily sentiment across multiple asset classes in one cheap subscription.
Strengths
- Very cheap entry (~€19.99/mo) for a bundled news + sentiment feed
- Two outputs: per-article sentiment (polarity + pos/neg/neu) and a separate daily-aggregated -1 to 1 Sentiment Data API
- Built-in ticker and topic tagging across stocks, ETFs, forex and crypto
- Sits inside the wider EODHD suite (EOD prices, fundamentals, calendars) under one key
Watch-outs
- News is delayed 15-60 minutes from publication, not for fast news trading
- Free tier is effectively unusable: each news call costs 5 base calls plus 5 per ticker, so ~2-4 news requests/day
- Lexicon-style polarity scoring with no published accuracy/benchmarks
- Modest, vaguely stated source breadth; unclear history depth and non-US coverage
Finnhub, News Sentiment API
Market data + news sentiment APIA clean per-company US sentiment roll-up (buzz, bullish/bearish %, sector baselines) bolted onto a full market-data stack, convenient, but US-only and not per-article.
- Sources
- Company news aggregated by Finnhub, Sector-average benchmarks across US companies, Separate /social-sentiment (Reddit/Twitter)
- Access
- REST API (JSON), Official SDKs (Python, Go, JS), WebSocket (for quotes, not sentiment)
- Best for
- Teams already using Finnhub's market data who want a cheap, ready-made per-ticker US news-sentiment score and buzz gauge for a dashboard.
Strengths
- Clean ticker-keyed roll-up: companyNewsScore, a nested bullish/bearish breakdown, weekly buzz, plus sector-average baselines
- Tight integration with a full market-data stack (quotes, fundamentals, company news, separate social-sentiment) under one key
- Generous free tier (60 calls/min) and well-maintained official SDKs
- Entity tagging is trivial: you query by symbol
Watch-outs
- News-sentiment endpoint is documented as US companies only
- Per-company aggregate, not per-article. You cannot trace which headlines drove the score
- Raw /company-news gives only headline, summary, source; companyNewsScore methodology is undocumented
- Generalist platform; production use beyond prototyping likely needs a paid plan
Polygon.io (now Massive)
Market data API with built-in news + sentimentPer-ticker sentiment labels with reasoning text, free on the Basic tier and backed by deep market data and long news history, but only a coarse categorical label updated hourly.
- Sources
- Multiple third-party news publishers in Ticker News, AI/LLM-generated sentiment label and reasoning per ticker, Separate Benzinga partner feed (no sentiment field)
- Access
- REST API (JSON), WebSocket streaming (market data), Official client libraries; flat files (higher tiers)
- Best for
- Developers wanting one vendor for deep US market data plus news with per-ticker sentiment labels, on a tight budget starting free.
Strengths
- Ticker News with sentiment is included on every Stocks plan, including the free Basic tier
- Sentiment is attached per-ticker inside each article's insights array with a human-readable sentiment_reasoning string
- Sits inside a deep, well-regarded market-data API (prices, aggregates, trades/quotes, options) under one key
- Long news history back to June 22, 2016 on paid plans for backtesting
Watch-outs
- Sentiment is a coarse categorical label with reasoning text, no numeric score, magnitude or confidence
- Ticker News updates only hourly across all plans; the real-time Benzinga feed has no built-in sentiment
- News/sentiment is a secondary feature; model, language and source breadth are thinly documented
- Free tier capped at 5 calls/min and 2 years of news history
Financial Modeling Prep (FMP)
Cheap news + sentiment APICheap ticker-tagged news inside a broad data API, but per-article sentiment lives only on a legacy v4 endpoint; the current stable news endpoints return no sentiment field.
- Sources
- Third-party publishers (e.g. Benzinga, Seeking Alpha), FMP's own articles/press releases
- Access
- REST API (JSON), Legacy sentiment RSS feed (JSON), Excel/Google Sheets add-on; API playground
- Best for
- Budget builders who already want a broad financial-data API and cheap tagged headlines to feed their own NLP, with an optional legacy sentiment label.
Strengths
- Very cheap entry ($19/mo) for news inside a large general-purpose API (fundamentals, prices, calendars)
- Clean JSON with one tagged ticker per article plus title, text, publisher and URL
- Legacy endpoint does return a Positive/Negative/Neutral label and 0-1 sentimentScore for a turnkey number
- Established vendor (since 2017, SOC 2 Type II)
Watch-outs
- Per-article sentiment lives only on a LEGACY v4 endpoint; current stable Stock News endpoints return NO sentiment field
- Sentiment methodology is opaque and reads like copy-paste from social sentiment (AHI/RHI 'hype index')
- Only one ticker tagged per item; no multi-entity tagging, topics or relevance scoring
- Free tier is EOD-only and excludes news; FMP frames the customer as the one doing the scoring
Benzinga News API
Editorial news API (sentiment is add-on/3rd-party)A broker-grade, well-tagged editorial newswire with WIIM catalysts and history to 2010, but the core feed has no native sentiment score, so you compute or buy it separately.
- Sources
- Benzinga in-house newsroom (original reporting), Press releases, news reports, SEC filings (for WIIM)
- Access
- REST API (JSON/XML), TCP push streaming, RSS feeds; WebSocket (wider suite), AWS Marketplace free Basic tier
- Best for
- Teams building entity-aware US/Canada news dashboards that want a clean editorial newswire and will compute sentiment themselves or layer a third-party dataset.
Strengths
- Excellent per-article entity tagging (symbol, ISIN, exchange) plus channels/tags and rich filters
- Original broker-grade editorial newswire with low-latency headlines and distinctive 'Why Is It Moving' catalysts
- Multiple delivery options: REST, TCP push, RSS, WebSocket; JSON/XML; archive back to 2010
- A genuinely free evaluation tier exists on AWS Marketplace
Watch-outs
- Core News API returns NO built-in sentiment, Benzinga's own guide runs TextBlob yourself
- Scored sentiment is fragmented (thin 'Newsquantified' docs; marketplace sentiment products are resold third-party datasets)
- Pricing is contact-sales with no published self-serve tiers; free tier is headline+teaser+link only
- US/Canada-centric; rate limits, latency and field schemas are not published
Tiingo News API
Cheap news API (no native sentiment)An extremely cheap, richly tagged US news feed from 120+ providers, but it returns no sentiment field at all, so you must run your own NLP.
- Sources
- 120+ integrated news providers and curated blogs, Algorithmic ticker and topic tagging by Tiingo
- Access
- REST API (query by tickers/tags/sources/dates), Community client libraries (Python tiingo, R riingo), Third-party integrations (QuantConnect)
- Best for
- Builders who want a very cheap, ticker-tagged raw news feed to run their own sentiment/NLP model on top of.
Strengths
- Extremely cheap: news included in the $30/mo Power plan, plus a free Starter tier
- Solid entity tagging using slang, company and product mentions, not just literal tickers
- Broad aggregation: 120+ providers across ~10,000 US equities (plus FX/crypto), history to Jan 2014
- Simple REST API with mature community libraries and QuantConnect integration
Watch-outs
- No built-in sentiment score, the decisive limitation for this use case; you must run your own NLP
- US-equity-centric, not a broad global multi-asset news graph
- Commercial use of news typically needs a separate license; pricing not transparent
- Queryable news-history window appears plan-gated and unclear
Where Adanos fits, and where it does not
Adanos lands mid-pack (4th of 12), in the upper half of the self-serve developer tier. It earns this on methodology transparency (a published hybrid VADER + Twitter-RoBERTa model with a -1 to +1 per-mention score), genuine source/publisher-breadth awareness, a fast 10-minute refresh, AI trend explanations, and raw article-level mention rows on Professional. It ranks below Marketaux and Alpha Vantage because those cover more asset classes (crypto and forex on top of equities) with comparably ready-made per-ticker scores, and below Stockdata only on raw breadth, not on transparency or reliability signals. It ranks above the 'no native sentiment' feeds (Tiingo, Benzinga core, FMP stable) and above Finnhub/Polygon on methodology clarity and refresh cadence.
Honest strengths
- Transparent, documented sentiment methodology (hybrid VADER + Twitter-RoBERTa, -1 to +1 per mention with pos/neg/neutral labels) where most rivals keep models opaque
- Source-aware by design: distinct publisher counts on trending, detail and market-sentiment endpoints, separating one-source noise from broad coverage
- Fast 10-minute refresh plus AI trend explanations and a clean compare/market-sentiment endpoint surface across 35,000+ tickers
- Raw article-level mention rows (snippet, source, timestamp) available on Professional for evidence behind aggregate scores
Honest limits
- Asset coverage is stocks and ETFs only, no native crypto or forex in this product, unlike Marketaux, Alpha Vantage, EODHD or Stockdata
- Scores are aggregated per ticker rather than offering multi-entity passage-level tagging within a single article (Marketaux and Stockdata do)
- Newer brand with a shorter public track record and retail/social-sentiment roots, not enterprise event-NLP heritage like RavenPack or MarketPsych
- Free and Hobby tiers are non-commercial and history caps at 365 days; commercial use and raw mentions require the $299/mo Professional plan, pricier than several rivals' top developer tiers
This dossier is published on Adanos's own website, so treat its placement of the Adanos Stock News Sentiment API as a self-assessment and verify the comparison against each vendor's live documentation.
FAQ
What is the difference between article-level and entity-level news sentiment?
Article-level (or full-body) sentiment scores the whole article, while entity-level sentiment assigns a score to each ticker mentioned. Many cheap APIs go further and only return a daily aggregate score per symbol. Marketaux and Stockdata score each tagged entity with passage-level highlights; Alpha Vantage returns both an overall and a per-ticker score; Finnhub and Adanos roll mentions up into a per-ticker score (Adanos also exposes raw per-mention rows on its Professional tier). Because methodologies differ, scores are not comparable across vendors.
Which financial news sentiment APIs have a genuinely usable free tier?
Marketaux (100 req/day, no card), Stockdata (100 req/day, no card), Alpha Vantage (25 req/day total), Finnhub (60 calls/min), Polygon/Massive Basic (sentiment included, 5 calls/min), and Adanos (250 req/month, non-commercial) all offer free access. EODHD's free tier allows only ~2-4 news requests/day, and StockNewsAPI has no permanent free tier (only a 5-day trial). Note many free tiers are non-commercial or heavily throttled.
Do these APIs cover crypto and forex news, or only US stocks?
Coverage is the sharpest dividing line. Marketaux, Alpha Vantage and EODHD tag stocks plus crypto and forex; Stockdata claims global multi-asset (though one review reports it narrowed to US-only). Finnhub's news-sentiment is US companies only, and StockNewsAPI, Benzinga, Tiingo and Polygon are US-oriented equity feeds; Adanos covers 35,000+ global stocks and ETFs, with crypto handled by separate Adanos products.
Which 'news APIs' do not actually include a sentiment score?
Tiingo returns tagged news with no sentiment field, Benzinga's core News API has no native score (its own guide runs TextBlob), and FMP's current stable Stock News endpoints return no sentiment, only a legacy v4 endpoint does. For these you must run your own NLP or buy a separate sentiment dataset.
How transparent are the sentiment methodologies?
Most vendors keep models proprietary. Alpha Vantage publishes its score thresholds, StockNewsAPI discloses it is keyword/title-based, and Adanos documents a hybrid VADER + Twitter-RoBERTa model. EODHD uses lexicon-style polarity, Polygon uses an undocumented AI label, and Finnhub's companyNewsScore formula is not published. Always treat scores as vendor-specific signals, not ground truth.
How far back does news sentiment history go for backtesting?
Deepest archives are Benzinga (to 2010) and Polygon/Massive (news to 2016 on paid plans). StockNewsAPI reaches 2019, FMP ~2019, Tiingo to 2014, and Alpha Vantage is confirmed back to at least 2022. Adanos caps history at 365 days (30/90/365 days by plan), and EODHD, Marketaux and Stockdata do not clearly publish their depth.
Sources
Primary sources checked while writing this piece. Vendor pricing and features change; figures reflect June 2026.
- Marketaux, marketaux.com
- Alpha Vantage News & Sentiment API, alphavantage.co
- Stockdata.org, stockdata.org
- Adanos Stock News Sentiment API, adanos.org
- StockNewsAPI.com, stocknewsapi.com
- EOD Historical Data (EODHD), eodhd.com
- Finnhub, News Sentiment API, finnhub.io
- Polygon.io (now Massive), polygon.io
- Financial Modeling Prep (FMP), site.financialmodelingprep.com
- Benzinga News API, benzinga.com
- Tiingo News API, tiingo.com