Dune

What it is

A SQL editor and dashboard tool sitting on a warehouse of blockchain data that Dune indexes and decodes for you. You write the query; what you are paying for is the indexing, the decoded schema and the execution engine behind it. The catalogue covers 100+ chains and, by Dune's own count, 3+ petabytes, organised into raw tables, ABI-decoded tables and curated cross-protocol tables such as dex.trades and nft.trades.

The corollary of buying a query engine rather than a metric set is that nothing is answered for you. There is no supplied definition of "active address" to accept or reject, no entity label turning an address into an exchange name, and no dashboard you did not build or borrow. The ceiling is your SQL and your willingness to learn a schema with a thousand tables in it.

Pricing

Four plans, all metered in one unit. Free gives 2,500 credits a month, Analyst $75 a month (or $65 billed annually, $780 a year) for 4,000, Plus $399 a month ($349 billed annually, $4,188 a year) for 25,000, and Enterprise is quoted. Legacy Premium, which had the cheapest credits on the board at $0.999 per 100, is closed to new subscriptions.

The API is included on every plan, the free one included. That is unusual in this category, where programmatic access is normally a separate and dearer product than the dashboard, and it is the main reason to look at Dune before the rest of the shelf. Dune's pricing FAQ says it plainly: all teams and users, including those on Free, can call the API to trigger executions and export data. Free-plan API keys get 15 requests a minute on write-heavy endpoints and 40 on read-heavy ones, which combine, so 55 a minute in practice.

What the free plan does not include is free execution. The free engine is query-editor only — "automated executions (API, scheduling, refreshing) are not available" — so every API call is a metered execution against those 2,500 credits. And the credit rate is where the plan price actually lands: past the allowance, Free pays $5.00 per 100 credits, Analyst $1.875 and Plus $1.596. The same query costs a free account about 2.7 times what it costs an Analyst account. Work out your monthly credits before reading the plan prices, because the gap between the rates is larger than the gap between the plans.

Credits reflect compute — CPU, bytes scanned, engine time — and Dune publishes no per-query formula, stating outright that costs appear after execution. Unused credits do not roll over, they cannot be bought in bulk, and the defences against a runaway bill are a per-execution cost cap and a monthly spend limit you set yourself. On exports, a credit returns 1,000 data points on Free and Analyst and 5,000 on Plus, where a data point is max(rows * columns, ceil(total_bytes / 100)).

One wrinkle for individuals: paid tiers are sold to teams, so a single user who wants Analyst has to create a team of one.

Data & coverage

Depth is uneven by design, and the differences are the reason to read the catalogue before committing. On EVM chains you get the full stack: raw transactions, logs, traces, creation_traces, blocks and withdrawals; ABI-decoded logs_decoded and traces_decoded plus per-contract call and event tables, which is contract-call level and not merely balances; and curated tables that span protocols, including dex.trades, dex_aggregator.trades, sandwich tables, NFT trades and a multichain stablecoin set with transfers and enriched balances for EVM, Solana and Tron.

Solana is a first-class citizen with its own shape — instruction_calls, account_activity, vote-transaction separation, IDL decoding submissions, dex_solana.trades and Jupiter aggregator swaps. Bitcoin is the other end of the range: blocks, inputs, outputs and transactions, with no decoded layer, because there is nothing to decode. Between them sit Aptos, Cardano (down to governance proposals and votes), Sui, Stellar, XRPL, Tron and Fuel, most of them raw plus token transfers rather than a full decoded stack.

Derivatives coverage is essentially Hyperliquid: raw, decoded and curated HyperCore data, with hyperliquid.market_data carrying order-book snapshots, funding rates, open interest and trade history. Third-party datasets ride along in the same catalogue — Flashbots' mempool dumpster, Farcaster via Neynar, Lens, Reservoir's NFT trading data.

What is not published anywhere in the documentation is a freshness figure. There is no stated lag between the chain tip and the catalogue tables, and no SLA for it outside an Enterprise agreement, so treat this as a warehouse rather than a feed and measure the lag yourself on the chain you care about.

Integrations

Official remote MCP server at api.dune.com/mcp/v1, authenticating by OAuth or an x-dune-api-key header, with tools for dataset discovery, query creation and execution, visualisations and dashboards. First-party SDKs the docs themselves name: dune-client for Python, @duneanalytics/client for TypeScript and duneapi-client-go for Go. Beyond the REST API there is a Trino connector for BI tools (the docs name Hex, Metabase and DBeaver) and a dbt connector with write access to DuneSQL. Scheduled queries can fire email or webhook notifications, which is the extent of alerting. The query editor has AI tools for generating, editing and debugging SQL.

Limitations

• The free tier's real constraint is the credit rate, not the credit count. At $5.00 per 100 credits past the allowance, free-plan overage is 2.7 times the Analyst rate and over three times Plus. Pay-as-you-go without upgrading is offered and is the expensive way to do it.
• Free executions do not exist over the API. The free engine is manual-editor only, so a free account automating anything is spending credits on every run.
• The Small engine times out at 120 seconds and allows three concurrent executions with no performance guarantee, which rules out the analysis you most want to run for free.
• No cost estimate before a query runs. Dune states there is no single per-query formula and that you see the cost after execution, so the only controls are the per-execution cap and the monthly spend limit.
• Credits expire monthly. No rollover, and no bulk purchase once the quota is gone.
• No entity labels. There is no first-party attribution layer turning an address into a named exchange or fund. Community-maintained tables exist inside the catalogue, but that is other users' work, not a maintained product with a revision date.
• No published data freshness. Nothing in the documentation states how far behind the chain the tables run.
• Results are capped at 32GB and are silently truncated past that unless you pass allow_partial_results=true, which turns the failure into a partial answer.
• Public dashboard URLs always carry a watermark, including for Enterprise teams.
• Every self-serve plan is internal use only. The vendor's own plan-comparison table has a "Data licensing and redistribution" row, and it reads Internal use only on Free, Analyst and Plus alike, with Custom commercial license reserved for Enterprise (read 2026-09-19). Price is not what separates the tiers here — the right to put the numbers in front of anybody outside your own organisation is, and it is not on sale at any published figure. Anything customer-facing built on a Plus plan is a sales conversation you have not had yet.

Alternatives

Within the same shape — a SQL editor over somebody else's index — Allium is the institutional counterpart: no free tier, no published price, but a first-party label set of 400M+ addresses and a stated freshness figure. The trade is the usual one at this end of the market. Dune is the only one of the two you can evaluate this afternoon without talking to anybody; Allium is the one with an answer when the question is who rather than what.

Footprint Analytics is the free alternative for people who would rather drag and drop than write SQL, over its own indexed warehouse — with a weaker public pulse and published coverage that contradicts itself, both of which that card documents.