NEAR has become the exchange everyone asked for but nobody shipped.
- Private and simple like a CEX, without the formalities. Send in one coin, take out another. Nothing public links the two, and nobody asks for your ID.
- Yours, and fair, like a DEX. It starts and ends in your own wallet. No account, and no house setting the price: solvers compete for your swap.
- What happens
- For
- Against
- Watch
Follow the fee
who payssettled by NEAR Intentswho is paid
Query
GET explorer.near-intents.org/api/v0/transactions -- every successful partner swap in the window -- left: fee dollars by what was swapped (sum(amountInUsd * appFees[].fee / 10000), grouped as in "Who's paying?") -- right: the same dollars by appFees[].recipient; the three NEAR revenue wallets are drawn in green -- band width is proportional to dollars; code: pipeline/swaps.py
The week it went dark
public readable on-chain
%#&$@ confidential
·
Query
-- NEAR's own saved query: dune.com/queries/5179085 (dataset dune.near.dataset_near_intents_metrics)
select date_at,
sum(case when source_type = 'public' then cast(volume_amount_usd as double) end) as public_volume_usd,
sum(case when source_type = 'confidential' then cast(volume_amount_usd as double) end) as confidential_volume_usd
from dune.near.dataset_near_intents_metrics
group by date_at
-- cross-check: api.llama.fi/summary/dexs/near-intents?dataType=dailyVolume
-- (equals public before 2026-09-01 and public + confidential from that day on)Who's paying?
;:+/| share of volumefees share of fees
Query
GET explorer.near-intents.org/api/v0/transactions
?statuses=SUCCESS,REFUNDED,FAILED&numberOfTransactions=1000
&startTimestampUnix={day start}&endTimestampUnix={cursor} -- paged back through each day
-- per successful swap: fee = amountInUsd * sum(appFees[].fee) / 10000
-- group: both sides in {USDC, USDT, DAI} -> stablecoin pair; else touches BTC -> Bitcoin;
-- else ZEC -> Zcash; else LTC / DOGE / BCH -> other UTXO coin; else everything else
-- code: pipeline/swaps.pyNobody opens a NEAR app
name share of partner volumeright: fee in bps · bps named to NEAR
Query
-- same swap records as "Who's paying?", grouped by the referral field -- fee charged = sum(amountInUsd * appFees[].fee / 10000) / sum(amountInUsd) -- named to NEAR = the part whose appFees[].recipient is one of -- buybacks.multisignature.near, fefundsadmin.sputnik-dao.near, 1csfundsadmin.sputnik-dao.near
The cut is rising
near what NEAR kept;:+/| the rest of gross fees
Query
GET api.llama.fi/summary/fees/near-intents?dataType=dailyFees -- gross fees, per day GET api.llama.fi/summary/fees/near-intents?dataType=dailyRevenue -- NEAR/wNEAR swept into its revenue wallets, per day -- summed by calendar month; share kept = revenue / fees. Full months only. -- DefiLlama builds both from NEAR's Dune datasets and the three revenue wallets (see "What the chain says").
So where's the money?
account partner's fee accountnear a NEAR revenue wallet
Query
-- same swap records, grouped by appFees[].recipient -- dollars = sum(amountInUsd * fee / 10000) per recipient, successful swaps only
From lunch money
fees partner fees per day
last 30 days
Query
-- three full days counted per month (the 5th, 15th and 25th), then the daily average of the window -- same Explorer request and fee rule as "Who's paying?"
The mix keeps rotating
btc Bitcoin%#& Zcashusd stablecoin pairs;:+ the rest
last 30 days
Query
-- share of each day's partner volume by group (groups as in "Who's paying?") -- three sampled days per month, so the path between them is not shown
It breaks sometimes
fail swaps refunded or failed, share of the day
Query
-- same Explorer request; per day: -- (count(status = REFUNDED) + count(status = FAILED)) / count(*)
The part we can count
swap partner swaps;:+/| the rest of official volume
Query
-- partner swaps: 2 * sum(amountInUsd) of successful swaps per day (Explorer API); -- doubled because the official figure counts both sides of a swap -- official: api.llama.fi/summary/dexs/near-intents?dataType=dailyVolume -- official swap counts, for the same check by count: dune.com/queries/4374089
The other side of the ledger
;:+/| official volume of the tokentoken partner swaps that touch itright: partner share · official volume
Query
-- official, per token and day: NEAR's saved query dune.com/queries/5157527 select date_at, symbol, sum(cast(volume_amount_usd as double)) as volume_usd from dune.near.dataset_near_intents_metrics group by date_at, symbol -- ours: sum(amountInUsd) of successful partner swaps whose origin or destination asset is the token -- (Explorer API, pipeline/swaps.py). Both count a swap once for each of its two tokens.
What the chain says
near NEAR arriving in the three revenue wallets, per day
Query
-- NEAR archival RPC, no key. For the last block of each UTC day:
query { request_type: "view_account", block_id: <block>, account_id: <wallet> } -- three wallets
block { block_id: <block> }.header.total_supply -- net new NEAR per day
-- wallets: fefundsadmin.sputnik-dao.near, 1csfundsadmin.sputnik-dao.near, buybacks.multisignature.near
-- code: pipeline/history.py the same thing in SQL: dune.com/queries/6740088 (a third party's)A meter nobody publishes
bytes new contract state per day
Query
-- NEAR archival RPC, no key. For the last block of each UTC day:
query { request_type: "view_account", block_id: <block>, account_id: "intents.near" }.storage_usage
-- the day's growth in bytes the contract stores: new accounts, keys, balances and intents
-- code: pipeline/history.pyWhat happened on those days
Charts jump on particular days. The middle column is measured by the build. The right column is what a public source says, with its link; we did not measure it, and where we found nothing we say so.
| Date | What our data shows | What was announced |
|---|
The neighbours
near NEAR Intents, one side of each swapvenue other venues, as DefiLlama reports them
Query
GET api.llama.fi/summary/dexs/{thorchain-dex, chainflip-amm, maya-protocol, near-intents}?dataType=dailyVolume
-- summed over the same days. NEAR Intents is halved, because its figure counts both sides of a swap
-- (see "The other side of the ledger"). The others are shown as reported; we did not audit their basis.What's the trade?
of new NEAR issued, in dollars
revenue a year · issued a year
all NEAR at today's price is a year of that revenue
Query
revenue = (revenue_now - cut_now * partner_fees) * growth
+ cut * partner_fees * growth
+ rest_bps / 10000 * (official_volume / 2 - partner_volume) * growth
issuance = net new NEAR in the window * price * (cap / 2.5%)
covered = revenue / issuance
-- run it yourself: python3 -m pipeline.model --cut 0.5 --growth 2 --cap 0.016
-- total_supply: RPC block(final).header.total_supply
-- max_inflation_rate: RPC EXPERIMENTAL_protocol_config
-- revenue_now: api.llama.fi/summary/fees/near-intents?dataType=dailyRevenue
-- wallets: RPC view_account now and 30 days earlier (archival node)
-- code: pipeline/model.py, pipeline/chain.pyThe bear case
Evidence against.
What would change our mind
Every line gets re-tested on each build.