AQUIX checked 2026-09-20

CheapTokens AI Inference

0x213c8d7434e2ae7aa1c392767c5120778d413215
#93 of 521 by verified x402 revenue top 18%| Unknown Declared · x402 directorywhere the name comes from
Most of what it received went back to the addresses that sent it — 89.8% returned to its own payers.

★ Whether any address is operated by an AI agent is not verified. On-chain records show that a payment settled, not who or what decided to make it.

What it says it does · published by the operator in a public x402 directory

OpenAI-compatible LLM inference and chat completions via discounted Venice API credits. Pay USDC to receive a live Venice.ai API key for /api/v1 models, chat, images, audio, video, and embeddings. Same-Day credits are 30–90% below Venice direct and expire at 23:59 UTC; Prepaid API Keys support recurring agent jobs.

Declared endpoint https://cheaptokens.ai/api/buy
Operator sitecheaptokens.aiobserved serving this address’s endpoints, not attested
Corroborated bycheaptokens.aithe operator’s own registered domain serves this address’s endpoints · a second independent signal
The operator’s claim, not our finding. We print it beside the money so you can judge whether the two agree. Against it, the chain shows $2,934.85 received from 42 payers across 941 payments, and $7.01 kept.
Measured
Received
$2,934.85USDC
lifetime, exact on-chain · 26.4% proven x402 by settlement record
Kept
≤ $7.01USDC
upper bound · USDC only
Payers
42
exact · every payer captured
Payments
941
exact total
Classification — what we decided from those measurements, never a rating
Verdict
Circular flow
behavioural evidence · no named rule was decisive · the base classifier decided
Evidence strength
Medium
42 payers · 301 distinct amounts · what this means
Identity
Name verified
what this tier means
Register status
In the register — circular flow
the other statuses

Verdict, every day we ran

Verdict each day
organic payments on all 28 measured days · current streak 16 days
Days we settled on chain
2026-08-112026-09-18
36 active of 39 days · a gap is a day with no settlement for this address, except the 3 marked mornings when AQUIX did not run · 11 outlined days were recovered from chain and carry no verdict
How a verdict can change
Every change is adjudicated in public and logged with its date and reason. Nothing here is edited silently, and no verdict is ever removed from this row.

Trends

$115.48$115.48$57.74$57.74$02026-08-11 in $45.782026-08-11 out $67.312026-08-12 in $38.592026-08-12 out $45.782026-08-13 in $18.872026-08-13 out $38.592026-08-14 in $25.662026-08-14 out $22.362026-08-15 in $22.112026-08-15 out $22.172026-08-16 in $10.842026-08-16 out $22.112026-08-17 in $22.142026-08-17 out $10.842026-08-18 in $16.752026-08-18 out $22.142026-08-19 in $9.472026-08-19 out $16.752026-08-20 in $29.54 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-20 out $0.002026-08-21 in $19.20 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-21 out $0.002026-08-22 in $9.79 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-22 out $0.002026-08-23 in $3.90 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-23 out $0.002026-08-24 in $26.97 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-24 out $0.002026-08-25 in $24.92 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-25 out $0.002026-08-26 in $106.93 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-26 out $0.002026-08-27 in $26.01 — SETTLED ONLY, recovered from chain. The all-route figure was never measured that morning, so this bar is lower than a normal day and is not comparable to one.2026-08-27 out $0.002026-08-28 in $25.182026-08-28 out $0.002026-08-29 in $14.512026-08-29 out $26.202026-08-30 in $34.772026-08-30 out $14.512026-08-31 in $30.472026-08-31 out $35.772026-09-01 in $1.002026-09-01 out $0.002026-09-02 in $49.612026-09-02 out $0.002026-09-03 in $3.042026-09-03 out $50.632026-09-04 in $46.792026-09-04 out $0.002026-09-05 in $54.642026-09-05 out $0.002026-09-06 in $71.212026-09-06 out $115.482026-09-07 in $71.912026-09-07 out $0.002026-09-08 in $50.042026-09-08 out $82.982026-09-09 in $34.482026-09-09 out $0.002026-09-10 in $77.472026-09-10 out $35.502026-09-11 in $49.192026-09-11 out $0.002026-09-12 in $40.912026-09-12 out $51.782026-09-13 in $70.922026-09-13 out $40.912026-09-14 in $66.052026-09-14 out $70.922026-09-15 in $60.182026-09-15 out $66.052026-09-16 in $47.822026-09-16 out $60.182026-09-17 in $37.282026-09-17 out $47.822026-09-18 in $55.302026-09-18 out $37.282026-09-19 in $54.452026-09-19 out $55.302026-09-20 in $7.012026-09-20 out $54.452026-08-112026-09-20
money in (above the line) money out (below the line) no row that day
All USDC routes, rail or not · 41 measured days between 2026-08-11 and 2026-09-20. In and out are opposite directions, drawn either side of zero and never summed. Kept, running total is received minus sent, accumulated, and floored at zero to match the Kept figure above · which is why a conduit's line sits flat on zero for the whole span: it forwarded everything. No per-day kept figure is stated, only the running total, because money can arrive on one day and leave on another.
Every day in the span as a table (39 rows, 36 measured)
daysettled USDCsettlementsaverage
2026-08-11$35.767$5.11
2026-08-12$38.514$9.63
2026-08-13$13.794$3.45
2026-08-14$20.586$3.43
2026-08-15$22.036$3.67
2026-08-16$1.202$0.60
2026-08-17$17.503$5.83
2026-08-18$16.7016$1.04
2026-08-19$9.414$2.35
2026-08-20$29.546$4.92
2026-08-21$19.2010$1.92
2026-08-22$9.799$1.09
2026-08-23$3.904$0.98
2026-08-24$26.9715$1.80
2026-08-25$24.9211$2.27
2026-08-26$106.937$15.28
2026-08-27$26.016$4.34
2026-08-28$26.1818$1.45
2026-08-29$9.4315$0.63
2026-08-30not measured — AQUIX did not run that day
2026-08-31not measured — AQUIX did not run that day
2026-09-01$83.078$10.38
2026-09-02not measured — AQUIX did not run that day
2026-09-03$3.003$1.00
2026-09-04$32.7718$1.82
2026-09-05$3.725$0.74
2026-09-06$15.907$2.27
2026-09-07$19.0010$1.90
2026-09-08$7.509$0.83
2026-09-09$25.4724$1.06
2026-09-10$29.1025$1.16
2026-09-11$21.6015$1.44
2026-09-12$14.2312$1.19
2026-09-13$5.056$0.84
2026-09-14$9.8510$0.99
2026-09-15$12.088$1.51
2026-09-16$12.357$1.76
2026-09-17$7.278$0.91
2026-09-18$13.2614$0.95
Read this chart with its coverage. Cumulative settled is 26.4% of what this address received; the rest arrived by paths the settled series does not reconstruct. It is not missing money — it is money this chart deliberately does not contain. Read it as evidence, not as a verdict. The outbound-per-day layer began recording on 2026-08-05 and is forward-only, so days before it were never measured.
36 measured days from 2026-08-11 to 2026-09-18. Drawn from settlements only · a day we did not measure is absent, never plotted as zero.

How it is trending

Every percentage is an average per measured day, never a window total. A total counts a morning AQUIX did not run as zero; an average simply has one fewer sample. A window needs 5 measured days on each side, and at least $1 a measured day to divide into, or it prints nothing at all.

Settled per measured day · 30d
+13.6%
$19.50 → $22.15
27 measured days against the 9 before them · window ends 2026-09-18
Settled per measured day · 7d
−39.4%
$17.47 → $10.58
7 measured days against the 7 before them · a short sample
Measured in total
$773.57
across 36 measured days
2026-08-11 to 2026-09-18

This address keeps 0.2% of what it receives. The percentage above is money moved, not money earned · a pipe getting busier is not a business getting richer. What it kept is under “What we read off the chain” below.

Where it sends

2destinations
0x8065…80ee90.0%
0x01ea…d7c910.0%
Shares of every USDC dollar that left this address, for its whole history. 2 destinations in total.

How it was paid

6sampled
carried a valid x402 authorisation83.3%
no authorisation we could read16.7%
x402 confirmed — 557 settlements on record. Every one is a USDC transfer whose transaction carried a matching EIP-3009 payment authorisation, reconstructed from public Base state. This is the proof, and it rests on all of them, not a sample. Separately, a sampler looked at 6 payments on 2026-09-19, 1 day ago, and classified them x402 direct. That is an indication from a small sample, not a finding about this address, and where it disagrees with the settlements above, the settlements are what count.

Where the money went

destination
USDC sent
share
0x8065…80ee
$2,635.06
90.0%
0x01ea…d7c9
$292.78
10.0%
These are addresses, not identities · we name one only where a public source already does. USDC only — the fold does not follow other tokens.

What we read off the chain measured

Facts, not judgements. Every figure below is USDC taken from Base settlements.

USDC received
$2,934.85
lifetime, all-route, exact on-chain
USDC sent on
$2,927.84
99.8% of received
Settled and attributed
$773.57
36 measured days
Distinct amounts
301
average $3.75
Volume processed · 30-day
$598.09
27 of 30 days measured · window ends 2026-09-18 · the rest are days AQUIX did not run
7 / 30-day trend
7d ▼ 39.4% · 30d ▲ 13.6%
average per MEASURED day, each window against the one before it · mornings AQUIX did not run are excluded, never counted as zero

What we concluded from them computed

Derived figures, each stating the arithmetic behind it.

Reality ratio
0.2389%
retained — processed
Confidence
Medium
42 payers · 301 distinct amounts
Volume backing
Low
$2,934.85 backed by 42 payers and 301 distinct amounts
Sybil state
clean
first-funder measured
Average per active day
$21.49
36 days with settlements
Consistency
92%
of days since first settlement had activity

Why this verdict

No named rule was decisive

None of our named rules was decisive for this address. Separately, 89.8% of everything it received went back to the addresses that sent it — it kept 10.2%.

Our rules that promote an address to a verified service refuse it above 30% circulation. Because no rule fired here that test was never applied — so rather than call it a service that earns, this page says what the money actually did.

This is not an accusation. Money returning to the same counterparties is exactly what a trading agent, a settlement loop or a refund flow looks like. We publish the measurement and the fact that no rule decided it; we do not infer intent from either.

89.8% of everything it received went back to the addresses that sent it.

Kept of received0.24%
Distinct payers42
Distinct amounts301

In, out and kept — 2026-09-19

One day, measured two ways. These are different bases: never add them together, and never divide one into the other. The day shown is always yesterday or earlier — the settled series is reconstructed the following morning, so today is never complete.
Received
Forwarded
Kept
Via the x402 rail
$6.42reconstructed from exact-scheme settlements
$0.00nothing paid onward on the rail
$6.42received — forwarded, both from the settlement store
All USDC routes
$54.45every USDC transfer in, rail or not
$55.30USDC only; the fold does not follow other tokens
$-0.86received — forwarded, both from the fold
Measured through Base block 51,534,080. One day only. There is no per-day series across the whole history: the outbound layer began recording on 2026-08-05, and this panel shows a single day from it.

Who funded the payers

Sybil stateclean
First funder of the payers0xf70d…dbef · hot-wallet, treated as infrastructure
Payers42 distinct paying addresses · 301 distinct payment sizes
A shared first-funder is a shared-funder pattern, so we name it. The test treats it as a signal only when the funder is not infrastructure and covers 50% or more of payers — most wallets legitimately get their first funds from one exchange, bridge or router.

Time and trust

First seen
Entered the register on 2026-08-11. Final through 2026-09-18 · 16-day streak.
Verdict history
Every snapshot ever taken, never rewritten. 28 snapshots recorded · no verdict change: this agent has held the same verdict since first seen.
Corrections
No published correction affects this agent.
Late writes
10 completed day(s), disclosed · a day finished after it was first published, and we say so rather than quietly restating it.
Basis
Snapshot 2026-09-20 · methodology v3.2. Retained and processed are never summed. Retained is clamped at zero, so it can overstate but never understate.
Check it yourself. Every figure on this page is reconstructed from public Base chain state. No API key, no account, no trust required — open the address and compare the USDC transfers against what we show.
Open on Basescan This page as JSON How we measure

Notes and limits

Paid through
coinbase 57 settlements $68.23 USDC 19 submitters · 2026-09-14 to 2026-09-20. 57 settlement(s) in the window. Every one of them is attributed to a named operator. Settlement-scoped and 7 days wide, a different basis from the lifetime total above. A submitter we cannot name is counted under unnamed, never dropped and never guessed.
Coverage of the daily series
26.36% of this agent’s money. The chart and the day strip are drawn from that series; the lifetime totals are not, and the two are never added together.
What this page cannot show for this agent
  • Which rule fired no named rule was decisive; the base classifier decided
34 of 35 items in our published contract are present for this agent; the full contract is in the JSON.
How a rule works here
One sentence per rule, stating what the rule tests and citing figures already on this page. The threshold itself is not published, so it cannot be gamed. Rules are named so that a verdict which changes can be pointed at. The full set of rules
Everything else is measured, not inferred. A figure we do not hold is never rendered as zero.
#93 of 521 by verified x402 revenue
AQUIX Evidence before trust @aquixai intelligence@aquix.ai
The record
All published agentsKnown, not measuredMethodology and rulesDefinitions
This page
PermalinkMachine-readable JSONWhy this verdictThe operator’s own site
Why you do not have to trust us

Every figure is reconstructed from public Base chain state. No API key, no account, no trust required — open an address and compare the USDC transfers against what we show. Every verdict is adjudicated in public and logged with its date and reason; nothing is edited silently and no verdict is removed from the record.

AQUIX · an independent observatory of the x402 rail on Base · Measured daily at 07:00 UTC · 2026-09-20 · No protocol pays to appear here

Company names and logos are the property of their owners and are shown for identification only. Their appearance here does not imply any affiliation with, or endorsement by, AQUIX.