Executive summary
The verified-live x402 count nearly tripled in May, from roughly 6,000 in April to roughly 18,000. On its own, that number tells a simple growth story. The structure underneath is more nuanced.
Most of the increase ran through a single host — Orbis — which positions itself as an API marketplace. Strip Orbis out and the market is still concentrated in crypto-native APIs, and new organic shipping decelerated even as the verified-live count climbed.
x402 verified-live nearly tripled month-over-month (~6k → ~18k).
Most of that growth was routed through one host (Orbis), now 70% of the visible pool.
Strip out Orbis and 55.64% of organic verified-live is Crypto/Blockchain. The market remains crypto-native.
Outside Orbis the verified-live count grew, but new monthly registrations fell 60%. Stock up, flow down.
For agents, what matters now is not how many x402 APIs exist, but which kind of supply they came from.
x402 nearly tripled
In May, the number of x402 APIs Decixa could actually call grew sharply. Counted as any API that returned a valid HTTP 402 at least once during the month, the working pool went from 6,162 in April to 17,929 in May. The chart below puts both months on the same scale, with the Orbis vs organic mix broken out.
x402 is no longer an experimental protocol with a handful of working endpoints. The number is real. But endpoint growth is heterogeneous — new independent providers, additional routes from existing providers, marketplace layers exposing thousands of slugs at once, listings that finally cleared verification, and templated/auto-generated endpoints all roll up to the same headline. The question to ask is not did x402 grow. It’s what kind of growth was it.
Orbis changed how the market looks
The single biggest structural change in May was Orbis. About 70% of the visible verified-live pool is routed through orbisapi.com:
This isn’t one provider building 12,000 distinct products. Orbis positions itself as an API marketplace, and what we see at the endpoint level is consistent with a marketplace/proxy layer rather than a conventional provider:
- URLs follow the pattern
orbisapi.com/proxy/<slug>-<6hex>. - API names are heavily templated and frequently auto-generated.
- Descriptions cluster around “Score X from Y”, “Calculate X from Y”, “Parse X from Y” phrasing.
- Explicit attribution to upstream data sources appears in under 0.3% of descriptions.
- Pricing is tightly clustered in a low fixed band.
- The functional surface frequently looks thin or generated rather than deeply differentiated.
None of this is a verdict on whether individual Orbis endpoints work — many of them do. It’s a structural observation: Orbis-mediated supply and organic provider supply have different shapes, are produced differently, and answer different questions. We separate them throughout this issue.
There is growth outside Orbis — but new shipping decelerated
Excluding Orbis, the verified-live pool still grew — from 3,283 in April to 5,277 in May (+60.7%). That number alone suggests organic shipping accelerated. The cohort view tells a different story. Cohort= the set of APIs newly registered in a given month, evaluated against today’s verified-live status.
Outer bar = registrations in that month. Filled = currently verified-live. Both cohorts share the same horizontal scale.
Total verified-live supply that exists at a point in time.
Number of new APIs registered during the month.
Share of registered APIs that actually reach verified-live.
In organic supply, stock went up, flow went down, and verification rate jumped sharply. The verified-live count grew not because more organic APIs shipped, but because the pipeline finally caught up on listings that had already been registered. Organic supply did not accelerate in May — it decelerated, with the surface obscured by a verification-rate windfall.
Without Orbis, x402 is still crypto-native
We classified all 17,176 currently-live endpoints into a bottom-up domain taxonomy. Then we re-ran the same aggregation on just the 5,128 endpoints that are not routed through Orbis. The headline taxonomy and the organic taxonomy disagree by a lot.
Read the second column on its own and x402 looks like a broad, general-purpose API marketplace. Read the third column and it looks like an infrastructure layer with crypto and blockchain at its core, surrounded by a few adjacent domains. We call that posture crypto-native — most of the diversity in the headline number turns out to be a function of one marketplace/proxy host.
Inside organic Crypto/Blockchain (n=2,853)
On-chain Data and Market Data together are 44% of organic crypto supply. That tells you something specific: today’s organic x402 crypto layer is read-oriented. It reads wallets, transactions, contract events, prices, volumes, DeFi metrics. Execution APIs — transaction submission, autonomous trading, agent-driven on-chain action — exist, but they aren’t the center of supply. x402 is becoming crypto-native infrastructure, but more for agents that read and decide than for agents that act.
Price formation is still early
x402 has a payment rail. Price formation, however, is still early. Both the median and the 25th percentile sit at $0.001 — a tenth of a cent — and about three quarters of all paid APIs are priced below one cent per call.
Note: pXX means the price below which that share of paid endpoints falls. p25 = lower quartile, p95 = top 5% boundary.
Median and p25 both at $0.001 strongly suggests price is being set as a default rather than as a function of value, compute cost, scarcity, or data quality. This isn’t a critique; it’s a description of an early market. x402 makes price machine-readable for agents, but price isn’t yet a meaningful signal. That changes when high-quality, reliable, scarce APIs start charging differently from the rest. We’ll watch the p95 band first.
What matters for agents
The biggest lesson from May is that endpoint counts are not enough to read the market. Visible supply nearly tripled, but most of that growth ran through one marketplace/proxy layer. Strip it out and the market is still crypto-native, and organic shipping decelerated.
For agents, what matters is not whether an x402 API exists, but where it came from, whether it actually works, whether the price is fair, whether it’s trustworthy, and whether it fits the task at hand. Without that, more endpoints do not mean more usable infrastructure.
The consequential question is no longer “how many x402 APIs are there?” The questions that matter are:
Which x402 APIs are actually usable for the task at hand?
Count is not coverage. 17,176 verified-live endpoints do not mean 17,176 distinct capabilities. Roughly 70% are routed through one marketplace/proxy host, and outside that host the market is overwhelmingly crypto-native.
Which APIs are trustworthy, fairly priced, and a fit for the task?
Verification status, cost, latency, and capability tags need to be legible side-by-side so an agent can pick autonomously. That’s the problem statement this report sits on top of.
Where is the gap between what agents need and what x402 supplies?
Outside Orbis, the non-crypto surface is thin and scattered across AI, Security, Communication, and Storage. APIs that let an agent act — not just read — remain the underserved slice.
Decixa is the search and decision layer for this question.
Methodology and caveats
Snapshot date: May 27, 2026. Verified-live refers to x402 endpoints that respond to an HTTP probe with a valid 402 handshake and a parseable payment requirement.
Three metrics, three questions. Stock = verified-live observed at a point in time (or any time in a month). Flow = APIs newly registered during a month. Verification rate = share of registered APIs that reached verified-live. These answer different questions and should not be compared interchangeably.
Treatment of Orbis.Orbis presents itself as an API marketplace. We separate Orbis-mediated supply from organic provider supply throughout this report because endpoint structure, naming, descriptions, and pricing differ materially between the two. This is not a quality judgment on individual Orbis endpoints — it’s a recognition that the two have different shapes and need to be reported as such.
Domain taxonomy. All 17,176 currently-live endpoints were classified into a bottom-up taxonomy from name plus description. Issue 01 (April 2026) used a k-means clustering over embeddings at k=9; Issue 02 uses a semantically-named 12-category taxonomy that re-runs cleanly month over month for comparability.
On the 5,523 number from Issue 01. Issue 01 reported 5,523 verified-live endpoints on an end-of-April snapshot basis. This issue uses 6,162 for April when comparing month-over-month, because it counts every API that returned a valid HTTP 402 at least once during April. Both are correct; they answer different questions. Where direct comparison is made, the basis is named explicitly.
Cost layer details
x402 prices today cluster heavily at $0.001 and $0.01— both default-looking values. Median and p25 are identical, which is a strong signal that the dominant pricing motion is “pick a default and ship,” not “price to value.”
What we’re watching in subsequent issues:
- Does the median move off $0.001?
- Does the p95 band widen?
- Do high-reliability or high-quality endpoints command premium pricing?
- Does price difference become meaningful enough for agents to use as a real decision signal?
What to watch in June
The Orbis trajectory
Does the 70% share keep climbing, stabilize, or compress as quality filters adjust? Do other marketplace/proxy layers emerge alongside it?
Organic shipping rate
May’s organic new-registration count fell 60%. Was that a transient dip, or structural deceleration? June’s cohort answers it.
Non-crypto organic growth
Outside crypto, Data & Search leads at 10.24% of organic supply. Whether AI/ML, Security, Communication, and Storage gain ground determines whether x402 expands beyond crypto-native infrastructure.
The write gap
Store and Communicate grew in absolute terms but remain well under 1% of supply. The leading indicator of x402 maturing beyond a read/analyze layer is when those categories cross into multi-percent territory.
Search the verified-live pool
All 17,176 endpoints in this report are searchable through Decixa. Pass natural-language intent, get back ranked endpoints with cost, latency, and verification metadata.