Benchmarks settled per call over x402. A free market-state reading, computed from US government statistics.
Current market state · Free feed
NEUTRAL
Distance from balanced: 0.03 / 1.00
Conditions as of
Mixed or balanced conditions. v3 · Magnitude, not a timing input. Methodology
x402 · Base mainnet
Pay-per-call APIs
2 services · USDC on Base · separate from the free feed. Prices follow the published specification.
Lending Benchmark
$0.02 USDC / call
Morpho lending practice, fees and utilization, read from contract state. Daily population statistics; no named vaults or curators.
GET /v1/lending-benchService details
Morpho Blue lending statistics computed from the protocol’s own contract state: liquidation-LTV practice by collateral class, curator fee norms, utilization, and how concentrated the market set is on shared oracle and interest-rate-model contracts. Coverage is every funded market reachable from the watched vaults’ withdraw queues, not every market on Morpho Blue, and the response reports that coverage in full. Bad debt and realised loss are not included — Morpho Blue exposes no bad-debt field, and the previous figures came from the venue’s own off-chain computation; the response names this exclusion and why. Daily population observations only; no curator, vault or market is identified.
Payer Census
$0.25 USDC / call
Who actually pays x402 sellers, read from Base. Real buyers separated from crawlers. Aggregates only — no seller and no wallet is identified.
GET /v1/censusService details
Every x402-style USDC settlement into every seller in the x402 catalogue over a trailing seven days, rebuilt from Base rather than from anyone’s reporting. A collection address receives USDC for many reasons, so a transfer is counted as a payment only when the same transaction also carries the EIP-3009 AuthorizationUsed event that x402 settles through; other inflow is reported separately rather than silently dropped. That makes this x402-style payments into x402 sellers, and it is stated as such rather than rounded up. Demand is then reported with crawlers removed: wallets paying twenty or more distinct sellers a day are machines checking endpoints, not buying from them, and for a long-tail seller they can be most of its apparent buyers. Reports totals, the buyer-versus-crawler split, how many listed sellers were paid at all, category mix, and how concentrated payment is among a few senders. No seller host, no collection address and no buyer or crawler wallet appears in the response.
Morpho · Ethereum & Base
Morpho Curator API
Track daily vault changes: fees, timelocks, curator replacements, and Morpho’s own vault warnings recorded as the venue published them. Recorded facts, not ratings or advice — a warning here is Morpho’s classification on a date, not an OpusAI assessment, and it is not derived from contract state.
regime-method.v3 · In effect from 24 September 2026
A description of observed market conditions, published as machine-readable JSON and served identically to every caller. It reports the state the market is in, not what anyone should do about it. There is no forecast, no recommendation and no consideration of any reader’s circumstances.
What it measures
The state is computed from two observed condition classes, both US government statistics:
Credit spreads — the high-yield option-adjusted spread, from FRED
Volatility term structure — spot VIX against three-month VIX, from FRED
Each is scored on a common scale where higher means calmer, then combined into a single measure of how strained conditions are. The two carry equal weight.
Every reading names its own provenance: sources gives what each component was read from, and input_as_of gives each series its observation date, which is not the date it was fetched — the two series publish on different schedules.
What it returns
regime — one of three words describing the state:
COMPRESSION — calm, benign, low-stress conditions
NEUTRAL — mixed or balanced conditions
STRESS — strained conditions, elevated volatility
state_intensity — 0 to 1, how far conditions sit from balanced. 0 is balanced; 1 is fully at one extreme. It is a magnitude, not a judgment about whether conditions are favourable, and not a timing input.
conditions — each input class reported coarsely: spreads TIGHT / NORMAL / WIDE, term structure NORMAL / FLAT / INVERTED.
as_of — when the underlying conditions were computed.
A new label must persist across several consecutive readings before it is published, so the series doesn’t flicker when a measure sits on a boundary.
When it returns nothing
The feed fails closed. It returns an error rather than stale or partial data when:
the published snapshot is more than 15 minutes old
the underlying conditions are more than 3 hours old
fewer than two input classes are available
a component cannot name the source it was read from
An outage here means the inputs were not sound. It never means old data was served as current.
These thresholds apply to the snapshot and the computation. Each series carries its own observation date in input_as_of, since the two publish on different schedules.
For developers
Paid x402 requests receive HTTP 402 with payment terms in the PAYMENT-REQUIRED header; settlement is in USDC on Base. No account is required. The free /v1/regime feed needs no payment and no key. Each reading carries sources and input_as_of alongside the state, so a caller can see what every component was read from and when the underlying series was observed.
Impersonal market data—not investment advice or an offer of advisory services. Outputs may be wrong, delayed or revised. Do not use them as the primary basis for investment decisions.
Full disclaimer
Not investment advice, not a recommendation to buy, sell, hold or avoid anything, and not an offer of advisory services. OpusAI, LLC is not an investment adviser under the Investment Advisers Act of 1940 or Florida Chapter 517 and accepts no fiduciary duty to users of this feed. Figures are model outputs, may be wrong, delayed or revised, and are not a representation of future results. Any use, including by software you control, is your responsibility.