Sentiment analysis APIs you can pay per call

What a sentiment analysis API returns

Sentiment analysis assigns an attitude to a piece of text. The simplest output is a label: positive, negative or neutral, sometimes with mixed as a fourth value. A score on a continuous scale, often from -1 to +1, adds intensity. Richer routes add a confidence value, the words that drove the result, emotions such as joy or anger, or sentiment toward a specific aspect, such as price or support in a product review.

Two families of routes share the name. The first scores the text you send: a review, a ticket, a post, a headline. The second chooses the text for you: it collects news about a ticker, company, topic or country and returns the tone of that coverage. The first measures your sample. The second measures a provider's sample of the media, which depends on its sources, its time window and how it matches your subject.

A third kind of number is often called sentiment too. Market indexes such as Fear & Greed combine prices, volatility, funding and other inputs into a mood reading for a market. They are useful in trading workflows and they do not analyze text you supply. This comparison lists them as adjacent products.

No score is a fact about the world. It is a model's or a rule set's reading of words, and it can miss sarcasm, negation, domain slang and context. Treat it as a measurement with an error rate you estimate on your own labeled sample.

Compare the returned data, sources and limits

The comparison separates two outputs that can qualify a service: a sentiment score or label for text you supply, and news or media sentiment about a company, topic or country. Market sentiment indexes, such as Fear & Greed or a crypto asset's mood from price action, and risk, regime or brief scores that use sentiment as one input are listed as adjacent outputs. They can accompany a qualifying service and never set the row price.

Zero-shot classifiers that sort text into your own labels can be used for sentiment buckets, and they stay out of the comparison: their contract is the label set you send, not a sentiment scale. Social attention rankings measure how much an asset is discussed, not how people feel about it.

A named language model is declared for the scoring by NetIntel, KR Crypto Intelligence and Stelar Digital. Model-based scores can handle context that keyword rules miss, and they can drift when the model changes, so keep the model name and the date with each stored result.

Crypto language or assets are named in the scoring by KR Crypto Intelligence, Stelar Digital and Onvexia. Domain tuning helps inside the domain and can hurt outside it. Korean-language news is named as the input by KR Crypto Intelligence. Emotion labels or emotion keyword hits besides polarity are returned by NetIntel, ThirdMade Shelf and AmanChain Relay.

Each price is the cheapest endpoint of that service for that kind of data: the same number wherever this page quotes it.

Compare 13 sentiment analysis services

Each row keeps the selected sentiment endpoint and its reference price. Rows are ordered by observed paying wallets for the service; the order does not rank accuracy, calibration or language coverage.

$/CALL is the price of the endpoint named under it: the cheapest sentiment score or label for text you supply and news or media sentiment about a company, topic or country endpoint of the service or, for a service without one, its cheapest sentiment endpoint. Never the cheapest tool of a gateway. Other kinds of data have their own price in the tables below. SENTIMENT ENDPOINTS counts the sentiment endpoints out of all the endpoints the service lists. BUYERS 30D counts distinct wallets that paid the whole service on-chain in 30 days, not only its sentiment endpoints. "Generalist" marks a gateway that sells more than three unrelated kinds of tools: it is never called the cheapest.

How to choose a sentiment endpoint

Build a small labeled sample first

Collect a few hundred examples from your own data and label them by hand: clear positives, clear negatives, neutral statements, mixed opinions, sarcasm and negation. Run each candidate on the same sample. Agreement with your labels is the number that matters, and it varies by domain far more than by price.

Decide what the scale means

A label, a score from -1 to +1 and a bullish or bearish signal are different scales. Decide how you will threshold a score, how you will treat mixed text and whether you need a confidence value to route uncertain cases to a person. Do not average labels from one provider with scores from another.

Match the domain

Customer reviews, support tickets, financial headlines and crypto posts use words differently. 'Bearish' is negative in a market comment and neutral in a wildlife article. Choose a route whose description fits your domain, and test it on that domain.

Separate your text from the provider's sample

If you need the tone of coverage about a company, a news route saves collection work. Ask which sources it reads, which time window it uses and how it matches the subject. If you already have the text, a text route gives you control over the sample.

Keep explanations as evidence

Routes that return the terms behind a score, a short reason or per-article results make errors visible. Store them with the score so a reviewer can see why a label was assigned.

Test the operating limits

The monitored median uptime is 100% and the median selected quote is $0.02. Monitoring latency describes our own checks, not the scoring time of a long text or a batch. Test long inputs, other languages and empty strings before committing a pipeline to a provider.

Choose by the output your workflow needs

Start with the answer your next step consumes. The candidates below use the same classification as the comparison table, and each job quote is the lowest priced endpoint of the requested types, with its path kept.

Score text you supply

A text route returns a label, a score or both for the words you send. Check the scale, how mixed or neutral text is reported and whether the route explains its score before you aggregate results across reviews, tickets or posts.

Candidates today, most buyers first: X Pay ($10.00/call, 351 buyers) · Content Intelligence API ($0.02/call, 312 buyers) · NetIntel ($0.002/call, generalist, 100 buyers) · AllTheBot ($0.003/call, generalist, 16 buyers) · ThirdMade Shelf ($0.001/call, generalist, 14 buyers) · Stelar Digital ($0.005/call, 13 buyers) · Onvexia ($0.002/call, 9 buyers) · AmanChain Relay ($0.007476/call, generalist, 3 buyers) · GigSoul x402 API ($0.04/call)

Follow news sentiment about a subject

A news route chooses the articles for you, so the answer depends on its sources, its time window and how it matches your subject. Keep the article list or the coverage window with the score.

Candidates today, most buyers first: KR Crypto Intelligence ($0.05/call, 84 buyers) · LoneStarOracle NewsSentiment ($0.05/call, 39 buyers) · Qonoro Intelligence API ($0.15/call, generalist, 14 buyers) · GDELT News Intelligence API ($0.02/call)

Read a market sentiment index

A market index is a published or computed number about trader mood, not an analysis of your text. Keep its definition and timestamp, since a Fear & Greed reading built from prices answers a different question from a scored headline.

Candidates today, most buyers first: Stelar Digital ($0.005/call, 13 buyers)

What the providers describe

These paragraphs attribute output contracts, models and limits to each provider. Reference prices follow the matching endpoint type, never an amount quoted in a marketing description.

Content Intelligence API

Content Intelligence API lists GET /content/analyze at $0.02. The provider describes one AI analysis that returns a sentiment score from -1.0 to 1.0, topic classification, named entities, a content type such as news, blog, docs or tutorial, and a credibility assessment. The service description adds that it extracts clean text, metadata and links from a URL. Sentiment is one field of a broader content profile here; if you need only polarity, compare what the extra fields are worth to your workflow.

1 sentiment endpoint of 2 · $0.02/call (GET /content/analyze) · Base/SOL · uptime 30d 100% · 786 ms · 312 buyers in 30 days

NetIntel

NetIntel lists POST /sentiment/analyze at $0.002 among its LLM-backed text tools. The endpoint description promises positive, negative, neutral or mixed polarity with a -1 to +1 score, emotion detection across eight labels such as joy, anger and trust, and aspect-based sentiment for customer feedback, via Claude Haiku. A model-based score can shift with phrasing and model updates; keep the model name and the observation date with stored results.

1 sentiment endpoint of 124 · $0.002/call (POST /sentiment/analyze) · Base/SOL · uptime 30d 100% · 107 ms · 100 buyers in 30 days

AllTheBot

AllTheBot lists POST /x402/sentiment at $0.003. The provider says the route classifies text as positive, negative, neutral or mixed with a -1 to +1 score, a confidence value and a short reason, in any language, as JSON. A reason field helps an analyst audit a label; it is the provider's explanation of its own output, not an independent check.

1 sentiment endpoint of 23 · $0.003/call (POST /x402/sentiment) · Base · uptime 30d 99.7% · 102 ms · 16 buyers in 30 days

Onvexia

Onvexia lists GET /v1/nlp/analyze and a batch variant, starting at $0.002. The provider says the scoring is tuned for crypto slang and tickers, because general models misread phrases such as 'going to zero' or 'wagmi', and that it returns polarity plus the terms that drove it. Domain tuning is a reason to test it on crypto posts and a reason to test it again before using it on product reviews.

2 sentiment endpoints of 57 · $0.002/call (GET /v1/nlp/analyze) · Base · uptime 30d 100% · 92 ms · 9 buyers in 30 days

Stelar Digital

Stelar Digital lists GET /sentiment at $0.005: a score from -1 bearish to +1 bullish and a label, either for arbitrary text or for a crypto asset based on its recent price action, via Claude Haiku. The two inputs answer different questions. A score of a headline measures its wording; a score from price action measures the market's recent move. Its market-brief bundles combine sentiment with regime, price and risk and are listed as derived products.

3 sentiment endpoints of 8 · $0.005/call (GET /sentiment) · Base/SOL · uptime 30d 100% · 362 ms · 13 buyers in 30 days

LoneStarOracle NewsSentiment

LoneStarOracle NewsSentiment has one route, GET /news, at $0.05. The service description says it scores recent headlines for a ticker, company or market topic and returns a bullish, bearish and neutral distribution with an aggregate score from -1 to +1. The headlines are chosen by the provider, so the score depends on how it matches your subject; a ticker that is also a common word needs a check.

1 sentiment endpoint of 1 · $0.05/call (GET /news) · Base/SOL · uptime 30d 100% · 994 ms · 39 buyers in 30 days

Qonoro Intelligence API

Qonoro lists POST /v1/news/sentiment at $0.15: news sentiment for a company with an overall score, signal strength, key themes, per-article sentiment and a recommended action for sales and PR agents. Per-article scores let you see which stories moved the aggregate. The recommended action is the provider's interpretation and should stay separate from the scores in any record you keep.

1 sentiment endpoint of 21 · $0.15/call (POST /v1/news/sentiment) · Base · uptime 30d 100% · 257 ms · 14 buyers in 30 days

GDELT News Intelligence API

GDELT News Intelligence API's media-tone routes start at $0.02. POST /news/sentiment tracks whether global media tone for a company, market, country or event is positive, neutral or negative, and whether it is rising or falling. An entity-graph route adds media tone over time with source-country distribution and the most active publishing domains. Media tone describes coverage; it is not a survey of public opinion.

2 sentiment endpoints of 5 · $0.02/call (POST /news/sentiment) · Base · uptime 30d 100% · 249 ms · 0 buyers in 30 days

KR Crypto Intelligence

KR Crypto Intelligence's news routes start at $0.05. The provider describes GET /api/v1/kr-sentiment as Korean crypto sentiment from an AI analysis that combines exchange data with Korean news. A semiconductor route returns English-translated Korean industry headlines with a Sonnet 4.6 synthesis that includes an overall sentiment and a bullish, bearish or neutral market signal. Translation is part of the pipeline, so tone can change between the Korean source and the English summary.

2 sentiment endpoints of 10 · $0.05/call (GET /api/v1/kr-sentiment) · Base/POL/SOL · uptime 30d 99.7% · 248 ms · 84 buyers in 30 days

ThirdMade Shelf

ThirdMade Shelf lists GET /probe/sentiment at $0.001: a positive, negative or neutral label plus emotion keyword hits. Keyword hits expose part of the evidence, but the description does not explain how the polarity classifier handles negation or sarcasm. Test both before choosing.

1 sentiment endpoint of 22 · $0.001/call (GET /probe/sentiment) · Base · uptime 30d 100% · 243 ms · 14 buyers in 30 days

Cost per text and for a larger stream

The median selected route costs $0.02 per call. At the same request shape, 100,000 calls total $2,000.00. The table repeats the calculation for the services named on this page; treat it as a reference figure for one text or one subject per call.

Each service at the price of its main sentiment endpoint, the one named in the table above.

Volume depends on how you slice the text. Scoring every sentence of a review costs more than scoring the review, and the answers differ. A batch route can score many texts in one paid call, which changes the arithmetic for a stream of short posts. News routes are billed per subject and per refresh, so match the refresh interval to how often the coverage changes.

Deduplicate identical texts before scoring them, and cache results for texts that do not change. Record whether an empty or failed analysis was charged under the provider's terms, and read the payment challenge for the exact request before authorizing a payment.

Inspect a request and its payment evidence

The public listing documents NetIntel at POST /sentiment/analyze, which scores the text in the request body. This endpoint has a reference quote of $0.002. Read the request schema before sending a body: the listing names the method and the path, not the JSON key that carries the text.

The listed method and description establish the advertised contract. This explanation does not reproduce an HTTP 402 challenge or certify a paid response. Open the linked service's payment evidence and request the challenge for the exact call before authorizing a payment. Keep the resource, method, input text, accepted payment and response together, then compare the label with your own reading of the text.

How this comparison is built

The comparison starts with services that are online and payment-ready, have a known price, a description of at least 80 characters and measured 30-day uptime of at least 90%. Task selection then reads the documented endpoints. A service enters the table only with a priced route that scores text you supply or the news coverage of a subject. Market indexes and scores that use sentiment as one input are shown as adjacent outputs. Zero-shot classifiers, social attention rankings, review velocity, news headlines without a documented sentiment field and research agents stay out, even when the service description mentions sentiment. Pages below the index threshold can still be read directly.

Provider claims were reviewed against the public service and endpoint descriptions captured on October 22, 2026, and remain provider claims. Monitoring supplies operating figures and reference prices; it does not test whether a label is correct. Buyer counts describe observed paying wallets for the service, with the coverage limits explained on its linked page.

The visible comparison date is October 26, 2026. Endpoint membership, row prices, the type table and the job candidates share one classification, so a cheaper market index cannot replace the quote of a text or news score.

Questions

How to pay

Each service answers an unpaid call with HTTP 402 and its price; your agent signs a USDC payment on the listed network and retries. No API key, no account, no subscription. See what x402 is.