Token Terminal

What it is

A curated set of financial metrics for blockchains and dapps, standardised so that two protocols can be read side by side and a protocol can be read against its own past. Token Terminal's frame is an income statement: fees at the top, supply-side revenue subtracted to leave revenue, then expenses and token incentives to leave earnings — plus treasury, TVL, daily/weekly/monthly users, trading volume, and quarterly and yearly financial statements.

Almost everything is daily. The catalogue's own table naming makes that explicit — one measure per table, per entity, per day — with the single exception of token price, which also carries an hourly table. Perpetual markets carry the deeper set: open interest, funding rate, liquidations, trader counts and position sizes.

Coverage is per metric, not per platform, and the gaps are wide. Read off the vendor's own metric pages on 19 September 2026: fees for 323 projects, revenue for 322, expenses 235, earnings 233, TVL 201, circulating market cap 718. A project appearing on the site does not mean it appears in the metric you came for.

Pricing

Free is $0 and carries full historical data, three custom dashboards and three custom charts. What else it carries is the one thing the pricing page will not say straight — see the end of this section.

Pro is $350/month, and that figure is the month-to-month rate: the Monthly/Annual switch is served unchecked, so $350 is what the page shows before anyone touches it, and annual billing is advertised only as "Save 15%" with no second number in the markup. It comes with a 7-day trial. What the comparison matrix gives it over Free is unlimited dashboards and charts, the Google Sheets and Excel plugins, CSV exports, team access, and a spreadsheet limit of 500,000 calls a month. Chat support is not one of the additions — the matrix ticks it in every column, Free included.

API has no published price — "custom pricing", via a demo booking. It is a separate tier rather than an add-on, so the $350 Pro plan does not include it. The published shape is 250,000 requests a day, a documented 1,000 requests a minute, the spreadsheet limit raised to 1M calls a month, Python and TypeScript SDKs, and an uptime status page.

Datashare is also custom-priced: raw and standardised tables shared into your own BigQuery or Snowflake project, with dedicated support and an SLA. The docs are blunt that query compute is billed to the caller's own Google Cloud project, which makes the subscription only one half of what this costs.

Payment is by card (Visa, Mastercard, Amex) or wire, team seats are billed per user per month, and there is a student rate by email.

The inconsistency worth knowing sits inside the pricing page itself. Its four plan cards list the Sheets and Excel plugin, access to MCP, CSV data exports and team access among the Free plan's bullets; its comparison matrix, further down the same page, leaves all four blank in the Free column and starts them at Pro. The other vendor pages split: the documentation FAQ sides with the matrix — data downloads need "an active Pro or API subscription" — while the MCP documentation sides with the plan card, saying the server is on every plan with limited usage on the free tier. So whether a free account can export a CSV or run the spreadsheet plugin is genuinely unsettled by Token Terminal's own pages, and this card does not pick a winner.

Data & coverage

Token Terminal runs its own nodes and works from transaction-level data rather than from other people's APIs — its own description is over a petabyte processed daily, loaded raw and transformed afterwards, so that a methodology change can be replayed over history instead of requiring a re-ingest. The stated benefit is auditability: every metric traces back to the source, and every metric definition is published per metric and again per project.

How many chains is the one number the vendor does not state consistently. The FAQ says they run their own infrastructure for "14+ blockchains"; the data-approach page claims standardised metrics across "hundreds of blockchains and thousands of applications". Individually verifiable on the explorer today: Ethereum, Solana, Base, Arbitrum One, Polygon, Bitcoin and Hyperliquid, among others.

The definitions are the product, and they are specific:

• Fees — the total users pay to use the application: borrow interest on a lending market, trading fees on a DEX, transaction fees on an L1 or L2.
• Revenue — the part of those fees the protocol keeps after its take rate, with the rest going to lenders, liquidity providers and other supply-side participants.
• Token incentives — native tokens paid out to users, contributors or investors, explicitly framed as the analogue of stock-based compensation.
• Earnings — revenue minus cost of revenue, token incentives and operating expenses, with the vendor's own caveat attached: it reflects on-chain profitability and may understate true costs where a project does not report its operating expenses on-chain.
• P/S ratio (circulating) — circulating market cap over annualised retained revenue.

Assets are modelled separately from projects, so a stablecoin or a tokenised fund carries supply, holders, mints, redemptions and peg deviation across chains rather than being flattened into one number.

Integrations

REST API at api.tokenterminal.com/v2 with a bearer key from the account page: projects, metrics, market sectors, assets, financial statements and prebuilt datasets including a crypto screener and an insider-transactions set. Rate limit 1,000 requests a minute, and a lapsed subscription answers 402 rather than 403, which is a useful distinction when something stops working.

The MCP server is hosted at mcp.tokenterminal.com/mcp, speaks streamable HTTP, and authenticates with OAuth 2.1 and PKCE — a browser login, no API key copied into a config file. It exposes one tool, research, and is available on every plan with tighter limits on the free one.

Google Sheets and Excel plugins cover the spreadsheet path; Datashare covers the warehouse one.

Limitations

• The API is quoted, not priced. No public number, no self-serve signup, a demo booking in the way. If the point of the evaluation was "can I afford to build on this", the pricing page will not answer it.
• The MCP server returns an opinion, not a dataset. The single research tool is described by the vendor as an AI-powered analyst over their data. It is useful for orientation and it is not a reproducible query: two runs of the same question need not agree, and nothing in the answer carries the provenance that the tables do.
• The multiples are an interpretation, and the vendor says so about one input. A P/S ratio inherits every judgement inside "revenue retained by the project", and earnings are published with a warning that off-chain costs are missing. These are valuation inputs, not valuations, and the catalogue's line holds here: this is standardised data, not a view on whether a token is cheap.
• Daily grain, with one exception. Every metrics table in the catalogue is _daily except metrics_tokens.price, which also has an hourly grain — and that is a Datashare table, not something the explorer or the REST API hands you. Intraday questions about fees, revenue, users or open interest are out of scope unless you buy the raw data and build them yourself.
• Coverage is uneven per metric, by a factor of three between market cap and earnings. Check the metric page before planning a screen around it.
• You cannot add a metric. The list is fixed; the only escape is Datashare, and Datashare charges you again in Google Cloud compute for every query you run.
• Listing is at the vendor's discretion. Projects are added weekly from submissions, with stated priority to submissions that arrive with complete smart contract descriptions.

Alternatives

DefiLlama covers an order of magnitude more protocols for free, with a published TVL methodology and a free API, but a looser accounting frame and no financial statements. Dune and Flipside answer the questions neither of them anticipated, at the cost of writing the SQL and owning the definition yourself.