Comparison·Updated July 30, 2026

Nexus vs Blockfrost: A Full-Stack Data API Compared to the Cardano Incumbent

Nexus or Blockfrost for your Cardano data layer?

If you build on Cardano, Blockfrost is probably the first API you ever called. It has been the default hosted gateway to the chain since 2020, it is open source top to bottom, and its OpenAPI spec is the closest thing Cardano has to a de facto data standard. Nexus is a younger, broader platform: the same on-chain reads, plus aggregated DEX market data, FIFO wallet P&L, server-side transaction building, a DEX aggregator, and WebSocket streaming, behind one REST surface.

The two products overlap on maybe a third of their endpoints and diverge completely on the rest, so this comparison goes layer by layer: API surface, architecture and consistency, rate limiting mechanics, pricing math, reliability posture, and the strategic question hanging over Blockfrost in mid-2026. Everything here was verified against primary sources in July 2026; links are inline so you can check our work.

TL;DR: Blockfrost is the better pure chain-data API: cheaper per request, deeper on core endpoints, 15 SDKs, and a real free tier. Nexus is the full-stack option: market data, wallet P&L, transaction building, streaming, and MCP on the same key, with rate limits that scale from 25 to 500 req/s while every Blockfrost tier below Enterprise shares a flat 10. Pure reads on a budget: stay on Blockfrost. Production apps that need throughput or anything beyond chain reads: Nexus. Facts verified July 2026. For the scannable version, see the side-by-side comparison; for the full Cardano surface, the Cardano API hub.

The July 2026 elephant: Blockfrost may stop being a company

Before the feature tables, you should know what is happening to Blockfrost right now, because it affects any long-term vendor decision.

There is a live on-chain governance action proposing to transform Blockfrost into a free, community-governed, not-for-profit public good. The treasury ask is about 9.8M ADA. The plan, per the proposal page: one public API free for everyone with no paid tier operated by the entity, commercial SLAs pushed out to independent partners, source code and trademarks transferred to a not-for-profit with a community-elected board, and a 99% monthly uptime commitment during the transition. IOG, which took over Blockfrost via a strategic investment in January 2024 (the site footer today reads "IOG Singapore Pte. Ltd."), would step back to a non-voting advisory role.

As of 2026-07-20 the proposal is polling far below the 66.67% threshold it needs (roughly 15% DRep yes by voting power on the Koios voting summary), with voting closing at the end of July. So the likely outcome is that it fails. But either result matters to you as a buyer: if it passes, Blockfrost's commercial tiers as they exist today go away; if it fails, you are buying from a vendor whose own owner just proposed to stop operating it commercially. Neither is disqualifying. Both belong in your risk model.

Nexus, for contrast, is a commercial product operated by Gero Labs (the GeroWallet team) with a conventional subscription model and no ownership questions in flight.

API surface, endpoint by endpoint

Both products are REST-first with a published OpenAPI spec (Blockfrost 0.1.90, June 2026; Nexus on OpenAPI 3.1, live at nexus.gerowallet.io). Blockfrost's merged spec contains about 128 paths; the Nexus docs cover about 130. But the composition is very different.

Three honest observations from that table.

Blockfrost is deeper where they overlap. On pure chain data, Blockfrost has more endpoints, more filters, and six years of accumulated edge-case handling. Its Conway governance group (19 endpoints covering DReps, proposals with CIP-129 addressing, and the constitutional committee, with additions as recent as June 2026) is substantially deeper than Nexus's current DRep set. It also has Cardano mempool endpoints, which Nexus does not.

Nexus is wider everywhere else. Market data, P&L, transaction construction, swap routing, and streaming simply do not exist in Blockfrost's spec, by design. A Blockfrost-based product that needs token prices typically adds TapTools or a custom indexer; needs P&L, builds FIFO accounting in-house; needs live prices, polls. Each of those is another vendor or another internal system to operate.

The multi-chain stories differ in kind. Blockfrost added hosted Midnight infrastructure in April 2026: a GraphQL indexer, GraphQL WebSocket subscriptions, and node JSON-RPC, all beta, bundled on every tier. Nexus serves Midnight over the same REST surface as its other chains, and goes further on the transaction side: DUST tracking, NIGHT/DUST registration builders, unshielded transaction building and submission. Nexus also runs live Bitcoin endpoints (self-hosted bitcoind plus an ord indexer for ordinals and runes) and Apex Fusion networks. Blockfrost has no Bitcoin story. One nuance to plan around: Nexus API keys are scoped per chain and network, so multi-chain means multiple keys under one account and one bill, not one key literally everywhere.

Architecture and the consistency question

Blockfrost's hosted API and its self-hostable twin (blockfrost-backend-ryo) are a Fastify service reading from cardano-db-sync's PostgreSQL, with Ogmios handling transaction evaluation. That architecture has one documented, still-open consequence: issue #86, filed by the cardano-transaction-lib team, describes how db-sync and Ogmios sit at different chain tips, so the UTxO view and the evaluation view can disagree. The maintainer's on-record position is that absolute consistency between the two tools is not achievable. In practice this bites chained-transaction flows: a transaction you just evaluated or submitted may not yet be visible to /addresses/{address}/utxos when you build the next one. The issue has been open since 2022 and db-sync remains the data layer, so it is still a real edge case in 2026, and client-side retry logic remains the standard workaround.

Nexus runs a different topology: a Java 21 / Spring Boot service in front of a prioritized provider chain per network. Reads are served by Nexus's self-hosted yaci-store indexers first, with Koios and, yes, Blockfrost itself as circuit-breakered fallbacks (Resilience4j per provider). The interesting part for this discussion is the health model: a tip-staleness check removes an indexer from routing when it is alive but serving stale data, rather than only when it is down. That does not make read-your-writes magically guaranteed across a distributed system, and we will not claim it does, but it is an architecture that treats "responding with old data" as a failure mode, which is precisely the failure mode behind issue #86. If chained transactions are your workload, test both against your flow during a trial; it is a 30-minute experiment.

Being transparent about that provider chain also tells you something else: Nexus depends on Blockfrost as a tertiary fallback for Cardano reads. We think disclosing that is more useful to you than pretending otherwise.

Rate limiting mechanics

The two systems throttle differently, and the difference matters at scale.

Blockfrost allows 10 requests/second per IP with a burst of 500, refilling at 10/s (a drained burst takes about 50 seconds to recover), on every tier; only Enterprise gets custom limits. Quotas are per day (50k on the free Starter, 300k on Hobby, 1M on Developer), reset at midnight UTC. The error ladder is specific: HTTP 429 for rate, 402 for exhausted daily quota, and 418 as an auto-ban if you keep hammering after a 429/402. One-day "boost" purchases add 300k or 1M requests when you blow the cap.

Nexus meters per key on two independent axes: a monthly request pool and a per-second token bucket whose depth is the burst. Both scale with tier: Builder gets 25 req/s with a 250-token burst, Growth 50/500, Scale 100/1,000, Enterprise 500/5,000, with per-key overrides available on Enterprise. Monthly-quota state is exposed on every metered response via X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset, and the two 429 variants are distinguishable (the monthly one carries the quota headers and a reset_at body; the per-second one just sends Retry-After: 1). Add-on and subscription gating uses HTTP 402 with a machine-readable addon field. Hot GET endpoints support ETag/304 revalidation, and revalidated responses still count against quota, so caching never silently changes your bill.

Structural differences to note: Blockfrost's paid tiers still share the flat 10 req/s ceiling, while Nexus's throughput grows with tier; Blockfrost resets daily, Nexus monthly; Blockfrost sells overage relief by the day, Nexus has no overage billing at all and simply returns 429 at the cap until you upgrade. Predictable, but a hard stop; size your tier accordingly.

Pricing math

Both vendors use flat subscription tiers, which makes them directly comparable, and honesty requires leading with this: on raw request volume per dollar, Blockfrost is cheaper, and it has a free tier while Nexus does not.

• Blockfrost (pricing): Starter is free forever at 50k requests/day (about 1.5M/month), one project. Hobby is €29/month for 300k/day (about 9M/month). Developer is €79/month for 1M/day (about 30M/month), with 15% off annually. Enterprise is custom. There is also a pay-as-you-go mode, payable in ADA natively or by card, and IPFS plus Filecoin storage is bundled into every tier.

• Nexus: Builder $29/month for 2M requests/month (3 keys, 7-day trial, the only trial mechanism; there is no free tier). Growth $79 for 10M. Scale $199 for 50M with streaming included. Enterprise $499 for 200M with every add-on and unlimited WebSocket bundled. Dedicated from $2,000 for isolated infrastructure. Add-ons: Market Data $19, Wallet Analytics $15, Transaction Builder $19, MCP $19, IPFS $5. WebSocket packages for Builder/Growth: $29, $59, $99.

So a maxed-out Nexus Builder with every add-on is about $106/month against Hobby's €29, for roughly a quarter of Blockfrost's monthly request volume. If your workload is on-chain reads and nothing else, Blockfrost wins that arithmetic outright, and you should use it.

The comparison inverts when you price the whole stack. The Blockfrost-era architecture for a trading tool or portfolio product is Blockfrost (chain reads) plus a market data vendor plus an in-house FIFO P&L pipeline plus a polling layer or custom streaming. The Nexus equivalent is one subscription where a request is a request across all of it. Whether $106 beats €29-plus-two-more-vendors-plus-an-indexer-on-call depends entirely on whether you need the upper layers. That is the actual decision, not the sticker prices.

One more Blockfrost pricing virtue worth naming: paying in ADA via pay-as-you-go is genuinely useful for DAOs and treasuries that hold ADA. Nexus bills conventionally (PayPal rails today) and offers a GERO-holder discount ladder instead.

Reliability posture

Blockfrost publishes a status page showing 100% component uptime over the trailing 90 days as of this writing, markets 99.9% uptime and 100M+ requests/day, and reserves contractual SLAs for Enterprise. That is a solid public record, and we will not pretend otherwise.

Nexus does not yet publish a public status page or a third-party uptime figure, and we will not invent one here. Its reliability story is architectural (the multi-provider fallback and tip-staleness routing described above) and contractual: SLAs exist where a written agreement exists, on Enterprise and Dedicated, with Dedicated adding physically isolated infrastructure (own cluster and database), a negotiated SLA, and a 4-hour support response over a dedicated channel. For everything below that, the honest instruction is: run the 7-day trial against your production traffic pattern and measure. The rate-limit headers make it easy to script.

When Blockfrost is the better fit

Being specific, because several of these are decisive:

• Your workload is pure on-chain reads. Blockfrost is cheaper per request, deeper on chain-data endpoints, and battle-tested. Nothing in Nexus's upper stack helps you if you never call it.

• You want a real free tier. 50k requests/day free, forever, no card. Nexus's 7-day trial is an evaluation window, not a home for a hobby project.

• You need deep Conway governance data or Cardano mempool access. Blockfrost's 19 governance endpoints and mempool group have no full Nexus equivalent today.

• You want to self-host or need an exit hatch. The entire Blockfrost stack is open source, with two self-host paths (the full RYO stack and the lighter Dolos-based Blockfrost Platform 1.0, which shipped in May 2026 with Hydra support). Nexus is a hosted product, full stop.

• You lean on SDK breadth or framework defaults. Blockfrost has first-party SDKs in 15+ languages (about five actively maintained, led by blockfrost-js at ~23k weekly downloads) and first-class provider slots in MeshJS and Lucid Evolution. Nexus currently ships an OpenAPI spec, a live docs playground, and an MCP server, but no official SDKs and no MeshJS/Lucid provider, so with those frameworks Blockfrost is the path of least resistance today.

• You need webhooks or push-based tx notifications and prefer them to managing a WebSocket connection.

When Nexus is the better fit

• Your product touches prices, P&L, swaps, or live data. That is the entire upper half of the surface table. One subscription replaces a two-or-three-vendor stack plus in-house accounting code.

• You want transaction building as an API. Six server-side builders returning unsigned CBOR (transfers, delegation, stake registration with optional reward withdrawal, DRep registration, vote delegation with Trezor-compatible certificate encoding, withdrawals), signed client-side via CIP-30; keys never leave your side. With Blockfrost you bring your own builder library.

• You want a swap feature without DEX integrations. The aggregator quotes and routes across 15 venues (live-verified July 2026: Minswap V1/V2, SundaeSwap V1/V3, WingRiders V1/V2, VyFi, Splash, TeddySwap, MuesliSwap, CSwap, Genius Yield, SaturnSwap, ShadowBook, SnekFun) and builds the swap server-side, non-custodially, at a 0.10% fee (0.05% for GERO holders). There is even an embeddable widget with a partner revenue share.

• You need streaming on Cardano. SockJS/STOMP topics for prices, per-asset feeds, and executed swaps, snapshot-then-updates, so no polling loop against your rate limit.

• You want UTXO chains beyond Cardano from one vendor, today. Bitcoin (with ordinals and runes), Midnight (with transaction building), and Apex Fusion are live now, alongside Cardano.

• Your compliance posture requires isolated infrastructure with a negotiated contractual SLA, which is the Dedicated tier's entire reason to exist.

Bottom line

Blockfrost is the better pure chain-data API: cheaper per request, deeper on core and governance endpoints, open source, self-hostable, and ubiquitous, with the caveat of an unresolved dual-tip consistency edge case and an unusual amount of 2026 strategic uncertainty. Nexus is the better platform the moment your requirements climb above raw chain reads, because the climb is where Blockfrost stops entirely.

Evaluate it against your own traffic with the 7-day Builder trial: get an API key at nexus.gerowallet.io, browse the full endpoint reference and live playground at nexus.gerowallet.io/docs, or talk to us about Enterprise and Dedicated via the support page. If you want the migration path itself, the Blockfrost alternative page walks it step by step.

Frequently asked questions

Can I run Nexus and Blockfrost side by side?

Yes, and it is the recommended migration path. Nexus keys come with a 7-day Builder trial, so you can mirror production traffic against both providers, compare responses and latencies, and cut over only when the numbers match. They are separate keys and separate base URLs; nothing about either API prevents dual-running.

Does Nexus have an SDK like Blockfrost does?

Blockfrost ships 15 official SDKs and is the default provider in most Cardano frameworks. Nexus ships one official TypeScript SDK, @adlabs/nexus on npm, typed against the live OpenAPI spec, plus a merged NexusProvider in Mesh and an open Lucid Evolution provider PR. Any other language can generate a typed client from the OpenAPI 3.1 spec.

What does the failed governance vote mean for choosing a provider?

The July 2026 proposal to convert Blockfrost into a treasury-funded public good did not pass, so Blockfrost continues as a commercial provider and nothing changes today. It also means the free public-good version of Blockfrost is not coming on any known timeline, so the honest way to evaluate both products is as the commercial services they are right now.

How do Nexus and Blockfrost handle going over quota?

Blockfrost's daily quotas hard-stop with 402 responses, and repeated overage can escalate to a temporary 418 auto-ban. Nexus returns 429 with X-RateLimit headers, a Retry-After, and the reset time in the response body, and a paying key is never banned for hitting its ceiling.