LunarCrush

What it is

LunarCrush measures what people post about an asset. It reads public posts from X, Reddit, YouTube, TikTok, Instagram and news sites, assigns each to a topic — a coin, a stock, a person, a brand — and keeps time series of how many posts mention it, how much engagement those posts drew, how many distinct accounts are posting, the engagement-weighted share of positive posts, and the topic's share of its category's conversation. Next to that it profiles the accounts themselves: followers, engagement and which topics each one moves. The vendor's counts are 4,000+ cryptocurrencies and 2,000+ stocks, with crypto and stock topics carrying price, volume and market cap alongside the social series.

One subscription covers every way in — the web app (Discover), a REST API, an MCP server and a CLI — and limits are shared across all of them, which is why this is one card rather than two. It sits under the news job in market data APIs because what a buyer is paying for is an alternative-data feed read from code; there is no chain data in it.

The two scores, and what they are. Galaxy Score and AltRank are composites the vendor computes from the same inputs it sells separately. Galaxy Score is a 0–100 figure blending social activity, engagement-weighted sentiment and market momentum, each normalised, and measured against the asset's own past; the vendor's bands call 70–100 strong and below 30 weak. AltRank is the same blend ranked across every tracked asset, where 1 is the highest at that moment. Both are included in every plan and neither is priced on its own. They are derived metrics — a summary of attention and price change — and should be read that way: the published method names the inputs but not the weights.

Pricing

Read off the pricing page on 4 October 2026. Individual is 90 USD a month, Builder 300 and Scale 900, with annual billing at 72, 240 and 720 USD a month — 864, 2,880 and 8,640 USD a year. Each plan can also be bought for a single day at 5, 15 or 45 USD, which is how the vendor suggests trying one, because paid plans have no free trial and payments are final. Enterprise is quoted and is the only plan with a trial.

The request limits are the useful numbers: 10 a minute and 2,000 a day on Individual, 100 and 20,000 on Builder, 500 and 100,000 on Scale, shared across API, MCP and CLI. A request over the limit returns a 429 and is not billed. Builder is where "every endpoint" and Custom Collections start, so Individual is not the whole API. On commercial use the pricing page says only that it is allowed within the limits of the plan, and sends anything larger to sales; Individual is described as being for research purposes and Scale for products in production.

Data & coverage

Daily history goes back to January 2020, per the pricing page. Social metrics are aggregated on a rolling 24-hour window and archived into hourly and daily buckets, and the time-series endpoint takes an interval or a start and end date with an hour or day bucket. The vendor calls this real-time; what a caller actually gets is those rolling aggregates over REST, with no stream, so the card says delayed.

The definitions matter more than in most market data, because the same word means different things across vendors. Here "mentions" is posts mentioning the topic in the last 24 hours after topic classification and spam filtering; "sentiment" is the percentage of positive posts weighted by engagement, so one viral post outweighs a hundred ignored ones; "social dominance" is share of voice within the topic's category, not across all of social media. None of these is comparable to a raw keyword count from a scraper, or to Santiment's social volume, which is built differently.

Integrations

The agent interface at the docs host returns Markdown by default and JSON or CSV with a format parameter, and documents topic summaries, topic time series, top posts, categories, creators and search. Authentication is a bearer token. The MCP server takes the same key or OAuth; an unauthenticated initialize against it returned 401 on 4 October 2026. There is a CLI, and the web app has metric alerts and first-mention alerts on paid plans.

Limitations

• The scores cannot be reproduced. Galaxy Score and AltRank have named inputs and no published weights, so a backtest or a research note built on them rests on a formula the vendor can change without notice. The component series — mentions, engagement, creators, sentiment — are the inspectable part.
• The marketing reads the numbers further than the data does. The API page lists "trading signals" as a use case and the metric glossary describes AltRank as useful for spotting rotations early. Those are the vendor's claims about what the data predicts, not part of the product, and nothing on the card relies on them.
• AI-written summaries sit beside the numbers. A topic summary carries a machine-written explanation of the price move — "likely driven by bullish sentiment" on Bitcoin on 4 October 2026. It is commentary attached to the data, and the full version is behind the subscription.
• Platform coverage is a dependency. Every metric depends on access to the underlying networks. A platform closing or repricing its own API changes what gets counted, and the series will carry that change rather than flag it.
• The free tier is ambiguous. On 4 October 2026 the pricing page gave the free account no API allowance, while the API product page advertised 500 free requests a day and an unlimited Pro tier that the pricing page does not list. Plan on the pricing page.

Alternatives

Santiment puts social volume and sentiment next to on-chain and developer metrics, with published definitions and a cheaper paid entry. For news rather than posts, Messari is the research-and-news feed in the same job.