# Solana PropAMMs: engineering decision memo

Prepared by @Benyorke (IMD agent #19). Research captured 2026-10-26 UTC. Public information only.

## Decision

Build the next version around a compact fair-value update, persistent depth/risk parameters, and aggregator adapters that quote the same rules used at settlement. Keep inventory and stale-price limits onchain. Prioritize asset coverage by whether the desk can observe and hedge the asset, then test whether additional router and wallet distribution produces profitable fills after landing costs. WOOFi is a comparator in this memo, with the same evidence requirements as other providers.

The useful competitive distinction is who controls pricing, inventory, and transaction landing. HumidiFi documents shared oracle updates; Flint documents a shared venue for independent makers; several other venues expose swaps without publishing their price model. Their reported volume establishes distribution, not profitability or implementation quality. **Engineering recommendation:** copy verifiable interfaces and cost-saving mechanisms, then measure them against the team's own public-market benchmarks; do not reverse-infer an undisclosed strategy from a volume rank.

## Reading this package

The provider CSV contains dated daily observations, calendar sums, published historical share statistics, and separately labelled dashboard snapshots. Filter `metric_window` and `data_status` before charting. The pair CSV contains numerical observations where obtainable and explicit unavailable-volume records elsewhere. Empty means unknown, never zero. `estimate=true` identifies rounded dashboard values; `estimate=false` means a value was copied or arithmetically derived from reported data, not independently certified against the ledger.

The required columns lead each CSV. Additional columns preserve window semantics, query identifiers, mint/pool identity where available, source flags, and restricted-denominator shares. This memo also supplies the question, method and limits that would otherwise belong in a README.

**Evidence labels:** Observed = a primary indexer's returned data or an author's own measurement; Vendor claim = provider documentation, not independent validation; Engineering inference/recommendation = this memo's conclusion; UNVERIFIED = insufficient evidence. A venue's undocumented implementation remains unknown even if an indexer labels it a propAMM. No audit, identity, asset absence, or security guarantee is inferred from silence.

## Comparable volume baseline and limitations

The common calendar comparison ends on 2026-10-24, the latest daily observation shared by the selected provider APIs at capture. Dates separated by `/` below include both full UTC calendar days. Daily records start at 2025-09-01 or the first subsequently available record; missing days are not synthesized. Values come from `totalDataChartBreakdown[unix_day].Solana`, summing its provider subkeys. In particular, Tessera's global `totalDataChart` is **not** a Solana-only series.

Calendar sums are deliberately recomputed from daily records instead of copying `total7d`/`total30d`. Those rolling summaries can have a different endpoint. Chain shares divide by the same dates in [DefiLlama's Solana overview API](https://api.llama.fi/overview/dexs/solana?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume), retrieved 2026-10-26. The chain denominator includes indexed spot swap legs; it is not unique user order notional, and multi-hop routes can contribute multiple venue fills. No aggregator volume is added to DEX volume.

`share_of_propamm_pct` is left empty for these current observations: the full active propAMM universe is not measured by one accessible source. `share_of_observed_cohort_pct` instead uses only HumidiFi, BisonFi, Tessera V, GoonFi, SolFi V2, Scorch, ZeroFi, AlphaQ, Aquifer and QuantumAMM on matching dates. That denominator omits smaller venues and is disclosed rather than silently promoted to total propAMM volume. Historical paper shares occupy separate rows and have a different pair population and statistic.

Pair amounts are less complete. GeckoTerminal pool snapshots cover HumidiFi and ZeroFi; CoinGecko tickers cover AlphaQ; a Blockworks table provides rounded Tessera amounts. None should be summed across sources into a chain total. Snapshot dates identify capture, not a fabricated calendar window: rolling end times are unavailable except ticker refresh timestamps. Full-provider pair shares remain empty; `share_of_observed_pairs_pct` divides only by returned pools/tickers. It does not claim exhaustive coverage. Quote symbols do not establish redemption rights; wrapped assets and stock-labelled tokens retain distinct names/mints where the API provides them.

## Starting leads: verification and reconciliation

- **Verified as the paper's claims:** [arXiv v1](https://arxiv.org/html/2609.38056v1#S4), published 2026-09-29: sample 2025-09-01/2026-08-31; Base launch 2025-10-30; Monad early April 2026; Ethereum excluded for builder-ordered updates. Sample scope is not a census.
- **Verified primary implementation lead:** [LambdaClass Factory](https://blog.lambdaclass.com/announcing-propamm-factory/), dated 2026-10-14, describes El Zorro, signed offchain updates, builder inclusion and the Priority Update Registry. The router already appears in its [2026-07-14 introduction](https://blog.lambdaclass.com/propamms-pushing-ethereum-to-the-next-level-of-finance/); treating the router itself as an October launch is outdated. The [2026-09-22 audit update](https://blog.lambdaclass.com/lambdaclass-propamm-router-audited-version-now-live/) documents inclusion-time requoting, Uniswap V3 fallback and recipient minimum-output checks. These are Ethereum design references, not evidence about Solana implementations.
- **Corrected date:** [Solana's Block Building Wars](https://app.blockworks.com/report/solana-s-block-building-wars) is dated **2026-02-17**, not 2026-10-14. Its leader-client markout sample is 2026-01-26/2026-02-09. Mean and median comparisons differ; a favorable median can hide adverse tail fills. These observations are not a current ranking of builder quality.
- **Denominator conflict resolved:** [TDMM's analysis](https://tdmm.io/insights/blog/top-crypto-market-makers/), dated 2026-09-26, reports five named venues at **26.1%** of **USD 75400000000** Solana DEX volume over the preceding **30 days**. This is a market maker's analysis of DefiLlama, not independent corroboration of the same dataset. It omits other propAMMs and uses all pairs. It is not comparable to the paper's SOL/USDC mean daily shares or this memo's later calendar sums. Do not average them.
- **Provider-list correction:** the six starting names are not exhaustive. The [Blockworks overview](https://app.blockworks.com/analytics/solana/solana-prop-amms), retrieved 2026-10-26, also reports activity for ZeroFi, AlphaQ, Quantum, Riptide, Taurusfi and Flint. Aquifer's inactive flag there conflicts with positive dated DefiLlama observations. Treat current liveness as source-dependent.

## Architecture evidence shared across the provider sections

[Bitquery's RFQ/propAMM documentation](https://docs.bitquery.io/docs/blockchain/Solana/solana-rfq-api/), retrieved 2026-10-26, separates **P: quote-posting** (HumidiFi, BisonFi, Tessera V, Aquifer, QuantumAMM, Scorch) from **O: oracle-at-swap** (SolFi, GoonFi, ZeroFi, AlphaQ). This is an instruction-traffic heuristic, not decoded pricing logic. Scorch uses a separate price program. AlphaQ's batched updates observed by Pine below demonstrate why O must not be read as “no updates exist.” The price source, update permissions, inventory skew formula and hedging stack remain UNVERIFIED unless separately sourced.

The same documentation warns that its trade-table coverage is incomplete historically and lacks ZeroFi rows. Indexer absence is not inactivity. The Solana program mapping below is from that primary indexer, retrieved 2026-10-26; program identity is a better join key than a symbol or token named after a venue.

| Provider | Swap program | Evidence class |
|---|---|---|
| HumidiFi | `9H6tua7jkLhdm3w8BvgpTn5LZNU7g4ZynDmCiNN3q6Rp` | P |
| BisonFi | `BiSoNHVpsVZW2F7rx2eQ59yQwKxzU5NvBcmKshCSUypi` | P |
| Tessera V | `TessVdML9pBGgG9yGks7o4HewRaXVAMuoVj4x83GLQH` | P |
| GoonFi V2 | `goonuddtQRrWqqn5nFyczVKaie28f3kDkHWkHtURSLE` | O |
| SolFi V2 | `SV2EYYJyRz2YhfXwXnhNAevDEui5Q6yrfyo13WtupPF` | O |
| Scorch | `SCoRcH8c2dpjvcJD6FiPbCSQyQgu3PcUAWj2Xxx3mqn` | P; separate updater |
| ZeroFi | `ZERor4xhbUycZ6gb9ntrhqscUcZmAbQDjEAtCf4hbZY` | O |
| AlphaQ | `ALPHAQmeA7bjrVuccPsYPiCvsi428SNwte66Srvs4pHA` | O heuristic; observed batching |
| Aquifer | `AQU1FRd7papthgdrwPTTq5JacJh8YtwEXaBfKU3bTz45` | P |
| QuantumAMM | `QuaNtZsgYRe5Z9Bk4LZ4cTD9tbkVoyCNf1R2BN9bBDv` | P |

## Fees, spreads and routing: common empirical reference

Observed SOL/USDC median realized round-trip spreads, **bps**, for **2025-09-01/2026-08-31**, published **2026-09-29**: [paper Table 1](https://arxiv.org/html/2609.38056v1#S7.T1). Selected matched fills; not fee tiers or causal estimates.

| Provider as named in paper | Aggregator flow | Other flow |
|---|---:|---:|
| HumidiFi | 0.09 | 0.86 |
| BisonFi | 0.71 | 1.01 |
| Tessera | 0.90 | 0.93 |
| SolFi V2 | 3.88 | 4.90 |
| GoonFi | 0.78 | 4.39 |
| ZeroFi | 1.52 | 1.50 |

Provider sections refer here rather than presenting these historical outcomes as today's contract settings. Router charges, wallet charges, priority fees, relay tips and the venue's embedded spread must be measured separately. **Recommendation:** a router identity alone must not grant a user a cheaper execution branch; prove which transaction context and authenticated counterparty were actually used.

## HumidiFi

### Architecture

**Vendor claim:** private pricing logic, offchain computation and onchain custody/accounting/settlement. The [current litepaper embedded on its homepage](https://www.humidifi.xyz/), retrieved 2026-10-26, describes independent pools sharing oracle infrastructure. The previously circulated `/litepaper` URL no longer resolves in this research; use the homepage text.

### How quotes are produced and updated

The vendor describes predictive inputs and inventory-aware quoting, with one oracle tick repricing supported markets. This supports sharing a fair-value anchor rather than rewriting every pool's book. Exact prediction model, state layout, staleness rules and live transaction costs are UNVERIFIED. It is class P in the indexer evidence table.

### Assets covered

Observed GeckoTerminal snapshots include SOL/USDC, cbBTC/USDC, WETH/USDC, PUMP/USDC, PENGU/USDC, TRUMP/USDC and Fartcoin/USDC. SNDK/USDC is stock-labelled; issuer backing is UNVERIFIED. The numerical pair CSV retains mint identities and other returned pairs. [Pool API](https://api.geckoterminal.com/api/v2/networks/solana/dexes/humidifi/pools?include=base_token,quote_token&page=1), captured 2026-10-26. LST activity was not established by this snapshot.

### Volume and share (windowed, sourced)

**Observed:** USD 39521743 on 2026-10-24; USD 283261380 over 2026-10-18/2026-10-24; USD 1066319584 over 2026-09-25/2026-10-24. The latter is 1.681408% of the same-dated indexed Solana DEX total and 4.952011% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/humidifi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Vendor lists Jupiter, DFlow, Titan and OKX Router in the homepage litepaper, retrieved 2026-10-26. A listed integration is not evidence of ongoing flow: the independent Blockworks aggregator snapshot discussed below shows a narrower distribution. Wallet access can be indirect through a router; direct wallet partnerships are UNVERIFIED.

### Fees and spread behaviour for aggregator vs non-aggregator flow

Vendor explicitly describes retail/adversarial-flow differentiation and says it adapts rather than blacklisting. Historical realized outcomes are in the common spread table. The current executable fee schedule, exact counterparty classifier and direct-flow availability are UNVERIFIED. The homepage's market-leadership percentages have no measurement window and are not used as current data.

### Known weaknesses

The pricing engine is undisclosed. Shared updates can create correlated stale-quote exposure across assets (**engineering inference**). Router availability and landing behavior are operational dependencies. This memo does not adopt the vendor's claims of congestion reduction or execution superiority without matched-fill tests.

## BisonFi

### Architecture

Class P. [DefiLlama's protocol methodology](https://defillama.com/protocol/bisonfi), retrieved 2026-10-26, identifies pool-state-discovered SPL/Token-2022 vaults. [Bitquery's dataset description](https://bitquery.io/datastore/datasets/bisonfi), retrieved 2026-10-26, independently identifies decoded swaps and reserve changes. Neither publishes the pricing curve. The [separately listed prediction-market program](https://tokenterminal.com/explorer/projects/bisonfi/about), identified by Token Terminal and retrieved 2026-10-26, must not be counted as this swap venue.

### How quotes are produced and updated

An operator publishes quote-related transactions, but the contents and trading model remain UNVERIFIED. The compact-update observation in the engineering cost table is a measured execution sample, not proof of a particular formula or of permanent ordering priority. Exact external reference venues and quote expiration are unknown.

### Assets covered

[Blockworks BisonFi](https://app.blockworks.com/analytics/bisonfi), retrieved 2026-10-26, lists SOL/USDC, SOL/USDT, USDT/USDC, cbBTC/USDC, whWETH/USDC, HYPE and TRUMP pairs. Its historical pair selector also contains equity-like symbols. Those labels do not prove present executable liquidity or identify the issuer. Current per-pair USD amounts were unavailable; corresponding major-pair cells are empty in the CSV. LST coverage is UNVERIFIED.

### Volume and share (windowed, sourced)

**Observed:** USD 219280910 on 2026-10-24; USD 2080852367 over 2026-10-18/2026-10-24; USD 8422622566 over 2026-09-25/2026-10-24. The latter is 13.281073% of the same-dated indexed Solana DEX total and 39.114836% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/bisonfi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed routed fills in [Blockworks' aggregator distribution table](https://app.blockworks.com/projects/solana/analytics/solana-dex-aggregators), retrieved 2026-10-26, establish Jupiter, DFlow, Titan and OKX paths. These are measured routes, not an API onboarding agreement or a direct wallet integration.

### Fees and spread behaviour for aggregator vs non-aggregator flow

See the historical spread table. A current public fee tier and whether its routing premium is an explicit fee, a classifier adjustment, or inventory pricing are UNVERIFIED. Do not treat the measured round-trip spread as revenue after hedging.

### Known weaknesses

Closed pricing and undisclosed update/withdrawal controls limit reproducibility. Reliance on aggregator flow makes route eligibility material. Engineering inference: stress tests should separate update landing failures from quote-model losses; aggregate volume cannot distinguish them.

## Tessera V

### Architecture

Class P on Solana. The [DefiLlama adapter](https://github.com/DefiLlama/dimension-adapters/blob/master/dexs/tessera/index.ts), retrieved 2026-10-26, explicitly separates Solana Dune trades from EVM event accounting. **Do not port the paper's decoded Base contract to Solana as fact:** its directional ladders are Base-specific evidence. [Paper §2.2](https://arxiv.org/html/2609.38056v1#S2.SS2), published 2026-09-29.

### How quotes are produced and updated

Solana update semantics, depth encoding and counterparty checks remain UNVERIFIED. Engineering inference: separate fair-price refreshes from less frequent depth changes, but benchmark that choice rather than assuming Solana matches the EVM implementation. The adapter's chain-specific accounting is also why the Solana breakdown is mandatory.

### Assets covered

[Blockworks Tessera Top Markets](https://app.blockworks.com/analytics/tessera), retrieved 2026-10-26, supplies SOL/USDC, SOL/USDT, cbBTC/USDC, whWETH/USDC and meme-pair amounts copied separately into the pair CSV. The Solana selector additionally lists SPCX, SPYx and JLP pairs; JLP is not a staking receipt. LST coverage is UNVERIFIED. Selector entries are historical coverage, not a current quote guarantee.

### Volume and share (windowed, sourced)

**Observed:** USD 135526658 on 2026-10-24; USD 1051807703 over 2026-10-18/2026-10-24; USD 3598293558 over 2026-09-25/2026-10-24. The latter is 5.673910% of the same-dated indexed Solana DEX total and 16.710551% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/tessera-v?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed Jupiter, DFlow, Titan and OKX paths in the Blockworks aggregator table, retrieved 2026-10-26. Multi-chain branding does not establish identical router access or wallet integrations on each deployment.

### Fees and spread behaviour for aggregator vs non-aggregator flow

The historical Solana result is in the common spread table. Current Solana fee fields, routing exemptions and quote-to-execution differences are UNVERIFIED. The paper's Base fee/spoofing findings are not a finding against the Solana program.

### Known weaknesses

Cross-chain totals and mechanism descriptions are easy to misattribute. Different source snapshots disagree in amount and endpoint; the CSV preserves rather than merges them. Engineering inference: require a deployment-specific adapter and quote replay corpus for each new chain.

## GoonFi

### Architecture

Class O heuristic. [DefiLlama's GoonFi methodology](https://defillama.com/protocol/goonfi), retrieved 2026-10-26, discovers markets and token vaults for both program generations. This memo uses the provider aggregate; it does not add a separate GoonFi V2 volume series and double-count the same provider.

### How quotes are produced and updated

The current pricing reference, inventory response and any external updater are UNVERIFIED. The absence of a visible quote flood in an indexer's heuristic does not establish a passive curve or a lack of offchain computation.

### Assets covered

SOL/USDC is supported by the historical measurement; [Blockworks' GoonFi dashboard](https://app.blockworks.com/analytics/goonfi), retrieved 2026-10-26, exposes a pair-volume query but no usable numerical export here. Current wrapped-asset, LST, meme and stock coverage remains UNVERIFIED unless a live adapter or trade trace establishes it. No unavailable pair volume is replaced by zero.

### Volume and share (windowed, sourced)

**Observed:** USD 45559492 on 2026-10-24; USD 409604349 over 2026-10-18/2026-10-24; USD 2179098168 over 2026-09-25/2026-10-24. The latter is 3.436075% of the same-dated indexed Solana DEX total and 10.119778% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/goonfi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed Jupiter, Titan and OKX fills in the Blockworks distribution table, retrieved 2026-10-26. Its blank DFlow cell is unknown in that snapshot, not proof that an integration is impossible.

### Fees and spread behaviour for aggregator vs non-aggregator flow

See the common spread table. The observed historical routing difference warrants testing direct, approved-router and nested-router paths. Current classifier rules and fee tiers are UNVERIFIED.

### Known weaknesses

DefiLlama records a swap-logic incident dated **2026-03-28**, reported loss **USD 254000**, on its [protocol page](https://defillama.com/protocol/goonfi), retrieved 2026-10-26. This is a primary indexer's incident entry; a provider postmortem and deployed remediation were not verified. Do not infer current vulnerability or safety from that entry alone.

## SolFi V2

### Architecture

Class O heuristic; distinguish its program from SolFi V1. The [DefiLlama summary](https://api.llama.fi/summary/dexs/solfi-v2?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume), retrieved 2026-10-26, identifies the version explicitly. [Blockworks Ellipsis](https://app.blockworks.com/analytics/ellipsis), retrieved 2026-10-26, attributes SolFi and Phoenix to Ellipsis Labs; Phoenix is a different venue.

### How quotes are produced and updated

Oracle-priced execution is documented by the indexer, while the model and update transport are UNVERIFIED. A public description grouping SolFi versions is not sufficient to infer the current version's oracle account format or fee implementation.

### Assets covered

SOL/USDC is established by the version-specific paper sample. The broader SolFi dashboard aggregates history across versions, so its other asset labels and stablecoin concentration are not assigned to V2 without program-level evidence. Current LST, wrapped BTC/ETH, meme and stock coverage by version remains UNVERIFIED.

### Volume and share (windowed, sourced)

**Observed:** USD 14447600 on 2026-10-24; USD 206143087 over 2026-10-18/2026-10-24; USD 981187417 over 2026-09-25/2026-10-24. The latter is 1.547169% of the same-dated indexed Solana DEX total and 4.556655% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/solfi-v2?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Blockworks reports Jupiter, DFlow, Titan and OKX flow to “SolFi,” retrieved 2026-10-26. **Version scope unresolved:** a provider-level observed route is not a version-specific integration certificate. The pair CSV explicitly labels V2 and leaves unsupported amounts empty.

### Fees and spread behaviour for aggregator vs non-aggregator flow

See the version-specific historical spread table. A live fee schedule and whether direct flow is permitted at the same price are UNVERIFIED. Never combine V1 and V2 spread or volume observations just because a dashboard displays “SolFi.”

### Known weaknesses

Version ambiguity can corrupt both coverage and routing analysis. Engineering inference: index by program ID and upgrade interval, and measure execution with the same context the router will invoke. The earliest indexer date is not necessarily a product launch date.

## Scorch

### Architecture

Class P with a distinct updater program (see shared evidence). The [protocol methodology](https://defillama.com/protocol/scorch), retrieved 2026-10-26, identifies token balances controlled by its authority PDA. Splitting price publishing from swap settlement is an observable architectural difference; the mathematical pricing function is UNVERIFIED.

### How quotes are produced and updated

Quote-related writes go through the separate price program. Counting only the swap program's non-swap instructions would miss those updates. The scope of a shared update, permissions and staleness enforcement are UNVERIFIED. Engineering inference: an indexer and landing monitor must follow both programs and the price accounts passed to swaps.

### Assets covered

The [DEXRabbit Scorch pairs table](https://dexrabbit.bitquery.io/solana/dex_market/scorch), retrieved 2026-10-26, displays SOL/USDC, SOL/USDT, USDT/USDC, cbBTC/USDC and PUMP pairs; recent trade rows also show BONK and PENGU. The page mixes window labels and conflicting totals, so its amounts are excluded from the quantitative tables. LST and stock coverage is UNVERIFIED.

### Volume and share (windowed, sourced)

**Observed:** USD 60974835 on 2026-10-24; USD 646218603 over 2026-10-18/2026-10-24; USD 2411918532 over 2026-09-25/2026-10-24. The latter is 3.803194% of the same-dated indexed Solana DEX total and 11.201000% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/scorch?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed Jupiter, DFlow, Titan and OKX routes in the Blockworks distribution table, retrieved 2026-10-26. A swap-program decoder alone does not establish correct price-state decoding for quoting.

### Fees and spread behaviour for aggregator vs non-aggregator flow

No matched aggregator-versus-direct spread measurement was obtained. **UNVERIFIED:** current fee tier, routing premium and default fee. DEXRabbit's generic fee estimate is not a measured Scorch fee and is not adopted.

### Known weaknesses

Separate update and swap attribution can cause incomplete cost accounting. The accessible pair dashboard's inconsistent labels make a verified pair export necessary. Engineering inference: freshness must be checked at the actual oracle account consumed by the fill, not the most recently seen transaction to the swap program.

## ZeroFi

### Architecture

Class O heuristic with program-owned inventory. The primary indexer notes missing trade-table coverage, so this report uses [GeckoTerminal pools](https://api.geckoterminal.com/api/v2/networks/solana/dexes/zerofi/pools?include=base_token,quote_token&page=1) and the DefiLlama daily series instead of reading missing Bitquery rows as no activity; retrieved 2026-10-26.

### How quotes are produced and updated

Price production, hedge inputs and update permissions are UNVERIFIED. Execution-time oracle classification cannot tell whether a reference is external, embedded in instruction data, or updated elsewhere; those are implementation questions requiring a trace.

### Assets covered

Observed numerical pool data covers SOL, USDT, cbBTC, WETH, PUMP, BONK, PENGU, JUP and other assets against USDC. Stock-labelled SPCX, MSTR, MU, TTWO, INTC and others are present; backing is UNVERIFIED. The pair CSV keeps distinct pool/mint identities and sums duplicate mint pairs only. [Additional pool page](https://api.geckoterminal.com/api/v2/networks/solana/dexes/zerofi/pools?include=base_token,quote_token&page=2), captured 2026-10-26. No LST coverage was established.

### Volume and share (windowed, sourced)

**Observed:** USD 28610010 on 2026-10-24; USD 190700649 over 2026-10-18/2026-10-24; USD 764304463 over 2026-09-25/2026-10-24. The latter is 1.205181% of the same-dated indexed Solana DEX total and 3.549446% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/zerofi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed Jupiter, DFlow, Titan and OKX fills in the Blockworks distribution table, retrieved 2026-10-26. Consumer wallet routing through these paths does not prove direct bilateral wallet integration.

### Fees and spread behaviour for aggregator vs non-aggregator flow

See the common spread table: the sampled routing categories do not establish a general non-aggregator premium here. Current fee settings and counterparty treatment are UNVERIFIED.

### Known weaknesses

Indexer coverage differs materially across sources. Engineering inference: breadth of listed pools should be checked against trade freshness and executable depth; a stock-labelled or zero-volume pool is not evidence that a market can be hedged or filled at meaningful size.

## AlphaQ

### Architecture

**Vendor claim:** oracle-driven pricing and dynamic fees using only its own assets, per [AlphaQ's website](https://www.alphaq.xyz/), retrieved 2026-10-26. It discloses proprietary trading rather than public deposit-based liquidity. The O classification is only a traffic heuristic and conflicts with a separate observed batch-update sample.

### How quotes are produced and updated

Pine's measured batched updates appear in the cost table below. This supports multi-market update amortization, but does not reveal the quote model. The maximum safe batch size, oracle validation and fail-closed behavior are UNVERIFIED.

### Assets covered

[CoinGecko's AlphaQ tickers API](https://api.coingecko.com/api/v3/exchanges/alphaq/tickers?page=1), captured 2026-10-26, supplies numerical USDT/USDC, JitoSOL/SOL, JupSOL/SOL, mSOL/SOL, whWBTC/cbBTC and SPYx/USDC observations, plus other stable/yield-token pairs. One JUPUSD/USDT ticker is source-flagged anomalous and retained with that flag. API mint strings are uppercased, so they are not copied as valid case-sensitive Solana mint addresses.

### Volume and share (windowed, sourced)

**Observed:** USD 21149410 on 2026-10-24; USD 200895102 over 2026-10-18/2026-10-24; USD 957932388 over 2026-09-25/2026-10-24. The latter is 1.510500% of the same-dated indexed Solana DEX total and 4.448658% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/alphaq?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed Jupiter, DFlow, Titan and OKX routes in the Blockworks distribution table, retrieved 2026-10-26. Neither this observation nor its website establishes exact onboarding requirements for a new router.

### Fees and spread behaviour for aggregator vs non-aggregator flow

**Vendor claim:** fees as low as **0.001%** on major pairs, equivalent to **0.1 bps**, [website](https://www.alphaq.xyz/), retrieved 2026-10-26. This is a minimum claim, not a universal fee or a measured router/direct split. Current pair-specific rates and spread discrimination are UNVERIFIED.

### Known weaknesses

Do not treat an O classifier as proof that batching is absent. Correlated-asset pools still carry redemption, depeg and issuer risk (**engineering inference**). The source-flagged ticker should be excluded from execution-quality conclusions until independently checked.

## Aquifer

### Architecture

Class P. Its [official frontend](https://aquiferdex.io/), retrieved 2026-10-26, exposed only a JavaScript shell in this research; no technical specification was verified. Operator identity, pricing curve and authority model remain UNVERIFIED.

### How quotes are produced and updated

An update stream is identified by the indexer and sampled by Pine. Exact price production, per-asset coupling and inventory control are UNVERIFIED. The measured update cost below is not an SDK or a promise of current cost.

### Assets covered

SOL/USDC is evidenced in the research sample. [Blockworks Aquifer](https://app.blockworks.com/analytics/aquifer), retrieved 2026-10-26, exposes a pair query but usable amounts were unavailable. A large historical catalogue cannot establish present liquidity. Current LST, wrapped BTC/ETH, meme and stock coverage is UNVERIFIED.

### Volume and share (windowed, sourced)

**Observed:** USD 11014298 on 2026-10-24; USD 88003063 over 2026-10-18/2026-10-24; USD 444089280 over 2026-09-25/2026-10-24. The latter is 0.700255% of the same-dated indexed Solana DEX total and 2.062360% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/aquifer?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Historical router presence is visible in Blockworks' Jupiter/aggregator dashboards, retrieved 2026-10-26; present route acceptance and direct wallet connections are UNVERIFIED. Do not equate historical selector membership with active route eligibility.

### Fees and spread behaviour for aggregator vs non-aggregator flow

No attributable routing-class spread or live fee schedule was obtained. UNVERIFIED, not assumed to match another provider.

### Known weaknesses

The positive DefiLlama daily series conflicts with the inactive flag in the Blockworks overview. Could reflect indexing scope, stale labels or a deployment change; cause is UNVERIFIED. Include its measured volume, but require a fresh successful swap and quote test before integrating it as live liquidity.

## QuantumAMM

### Architecture

Class P; “Quantum” in Blockworks and “QuantumAMM” in DefiLlama refer to the venue family represented by the program mapping above. The mapping is an indexer attribution, not a verified legal/operator identity. Its proprietary pricing formula is UNVERIFIED.

### How quotes are produced and updated

Quote publishing is observed. Pine's cost/cadence sample below shows why low per-update compute and low total network cost are different measurements. Exact feed aggregation, quote validity and controls remain UNVERIFIED.

### Assets covered

SOL/USDC is evidenced by the overview and paper sample. [Quantum's dashboard](https://app.blockworks.com/projects/quantum-propamm/analytics/quantum-overview), retrieved 2026-10-26, offers pair categories but no usable numeric export here. Other requested asset classes remain UNVERIFIED; no zero is inferred.

### Volume and share (windowed, sourced)

**Observed:** USD 4146420 on 2026-10-24; USD 56289714 over 2026-10-18/2026-10-24; USD 707297793 over 2026-09-25/2026-10-24. The latter is 1.115291% of the same-dated indexed Solana DEX total and 3.284706% of the observed cohort, **not** all propAMMs. [DefiLlama API](https://api.llama.fi/summary/dexs/quantumamm?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume); retrieved 2026-10-26.

### Aggregator integrations

Observed Jupiter, Titan and OKX routing in Blockworks' distribution table, retrieved 2026-10-26. Its DFlow cell is empty in that snapshot; support is unresolved.

### Fees and spread behaviour for aggregator vs non-aggregator flow

Current venue fees, embedded spread and routing-class differences are UNVERIFIED. The lack of a published matched-flow comparison is not evidence that it treats all counterparties equally.

### Known weaknesses

Offchain model and operational authority are undocumented in the examined primary sources. Engineering inference: measure update success, quote age at fill, aggregate cost and maker markout together; an update flood alone does not show that usable quotes remained fresh.

## Riptide

### Architecture

Blockworks classifies a traded Solana venue under this name; its public technical implementation is UNVERIFIED. Do not conflate it with similarly named launchpads, terminals or lending protocols. The dashboard attribution is adequate to retain a volume observation, not to infer custody or operator identity.

### How quotes are produced and updated

UNVERIFIED: price model, update transaction, oracle source, quote expiry and inventory policy. No public specification was reliably joined to the venue identifier.

### Assets covered

The Blockworks overview identifies NVDAx/USDC as its top pair. That supports investigating tokenized-equity flow, but mint-level identity, backing and depth are UNVERIFIED. Other requested asset coverage is unresolved.

### Volume and share (windowed, sourced)

**Observed rounded snapshot:** USD **6180000** over the dashboard's past **1 day**, USD **33300000** over **7 days**, retrieved **2026-10-26** from the [Blockworks overview](https://app.blockworks.com/analytics/solana/solana-prop-amms), chart `9177`, query `4687`. Exact window endpoints and comparable shares are unknown. These are not aligned to the calendar baseline.

### Aggregator integrations

Observed Jupiter, Titan and OKX routes in the Blockworks distribution snapshot, retrieved 2026-10-26. Direct wallet access and current integration agreements are UNVERIFIED.

### Fees and spread behaviour for aggregator vs non-aggregator flow

UNVERIFIED: no attributable fee tier or aggregator/direct spread comparison obtained.

### Known weaknesses

Provider identity and market identifiers require confirmation before integration. Engineering inference: equity-session closure, reference-price discontinuities and redemption constraints need explicit risk handling rather than an always-open native-token pricing policy.

## Taurusfi

### Architecture

A Solana venue with reported trading in Blockworks. Public technical specification, relationship to similarly named EVM venues, and operator identity are UNVERIFIED. A shared name is not proof of shared code.

### How quotes are produced and updated

UNVERIFIED: pricing inputs, update packaging, replay protection, stale-price policy and inventory management.

### Assets covered

Blockworks identifies SPCX/USDC as the top pair. The underlying instrument, issuance mechanism and executable depth are UNVERIFIED. Preserve the source symbol rather than silently reclassifying it as ordinary US equity shares.

### Volume and share (windowed, sourced)

**Observed rounded snapshot:** USD **95400** over the past **1 day**, USD **871000** over **7 days**, [Blockworks overview](https://app.blockworks.com/analytics/solana/solana-prop-amms), retrieved **2026-10-26**, chart `9177`, query `4687`. Endpoints and comparable shares are unknown.

### Aggregator integrations

The overview identifies OKX as its leading aggregator in that snapshot, retrieved 2026-10-26. A complete live integration matrix was not obtained; Jupiter/DFlow/Titan availability remains UNVERIFIED.

### Fees and spread behaviour for aggregator vs non-aggregator flow

UNVERIFIED: no live fee schedule or routing-class spread measurement obtained.

### Known weaknesses

Small reported activity plus undocumented mechanics does not establish a general-purpose asset platform. Engineering inference: do not copy an apparent stock focus without independently validating the reference/hedge path and session limits.

## Flint

### Architecture

**Vendor documentation:** a multi-maker onchain venue merging independently quoted liquidity into a virtual book with pro-rata fills at equal price levels. This differs from the single-operator model. [How Flint works](https://docs.flintlabs.dev/overview/exchange), retrieved 2026-10-26. Maker pricing and inventory decisions remain external to the shared execution service.

### How quotes are produced and updated

Documented alternatives are explicit OrderList, OracleOffset and LinearDistribution. Fair values can move a persistent ladder; sequence counters address replay. [Quoting documentation](https://docs.flintlabs.dev/sdks/quoting), retrieved 2026-10-26, describes atomic two-sided fair updates, slot-age checks and batch packing. Receipt acceptance and landing are distinct; do not treat gateway acceptance as execution.

### Assets covered

The [official site](https://flint.trade/), retrieved 2026-10-26, describes USDC-quoted spot. The Blockworks overview lists USDT/USDC as a top pair. Exact live native-token, LST, wrapped BTC/ETH, meme and stock inventory should be read from the current catalogue; coverage was not fully retrieved here. Future perps documentation is not evidence of live spot markets.

### Volume and share (windowed, sourced)

**Observed rounded snapshot:** USD **3180** over the past **1 day**, USD **17500** over **7 days**, [Blockworks overview](https://app.blockworks.com/analytics/solana/solana-prop-amms), retrieved **2026-10-26**, chart `9177`, query `4687`. The model is informative, but these amounts do not demonstrate comparable scale to the calendar leaders.

### Aggregator integrations

The official site names aggregator distribution; Blockworks' overview identifies Titan activity. Exact current Jupiter/DFlow/OKX support and onboarding status are UNVERIFIED in the primary sources inspected. Do not substitute earlier launch announcements for a live route test.

### Fees and spread behaviour for aggregator vs non-aggregator flow

Pine's catalogue pull dated **2026-08-24** reports **0.01 bps** stable/stable, **0.5 bps** major, **1 bps** volatile-pair venue fees: [original research](https://pineanalytics.substack.com/p/who-flint-wins-for), published **2026-08-25**. These are historical venue fees, not the maker's total spread or an aggregator/direct comparison. Current rates are UNVERIFIED.

### Known weaknesses

Shared landing infrastructure can correlate maker outages (**engineering inference**). The quoting docs say post-only is enforced during a later crossed-book sweep, not by rejecting crossing quotes at submission. Treating it as CEX-style immediate protection would be an integration error. Pluggable maker strategies also create shared matching and rounding complexity.

## WOOFi sPMM (comparator)

### Architecture

**Vendor documentation:** token-specific WooPool accounting/vaults, WooOracle pricing and administrative roles. The [Solana architecture guide](https://learn.woo.org/woofi-docs/woofi-dev-docs/guides/integrate-woofi-as-a-liquidity-source-on-solana/architecture), retrieved 2026-10-26, describes Pyth validation of WOO oracle values and separate configuration authorities. It is an oracle-priced design; do not assign the undocumented internals of another propAMM to it.

### How quotes are produced and updated

The guide describes an onchain oracle simulating external order-book conditions. WOOFi's [general sPMM math](https://learn.woo.org/woofi-docs/woofi-dev-docs/resources/the-math-behind-spmm), retrieved 2026-10-26, separates mid-price, spread and liquidity coefficient, with synthetic price impact dependent on notional. That is a published continuous-impact model rather than an explicit order ladder; this older general description does not verify the current Solana implementation. Deployed cadence, batching, compute cost, thresholds and hedge inputs remain UNVERIFIED. **Recommendation:** benchmark the Solana program; compact parameter updates need not require replacing the existing pricing model.

### Assets covered

The documentation supports per-mint pools but is not a current active-market list. Coverage across SOL, stables, LSTs, wrapped BTC/ETH, memes and stocks remains UNVERIFIED in this package; a configurable pool mechanism is not evidence that every category is live. No fictional pair-volume rows are added.

### Volume and share (windowed, sourced)

The [DefiLlama Solana overview API](https://api.llama.fi/overview/dexs/solana?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume), retrieved **2026-10-26**, explicitly returns **USD 0** in its **24h**, **7d** and **30d** WOOFi Swap fields. Those source zeros are retained with `reported_zero_activity_unverified`; they are not a claim of no onchain trading. No nonzero comparable Solana volume or share was verified.

### Aggregator integrations

The [official Solana integration guide](https://learn.woo.org/woofi-docs/woofi-dev-docs/guides/integrate-woofi-as-a-liquidity-source-on-solana), retrieved 2026-10-26, links Rust/TypeScript SDKs and a Jupiter integration. This establishes integration code availability, not current route volume. Other aggregator and wallet connections are UNVERIFIED here.

### Fees and spread behaviour for aggregator vs non-aggregator flow

The architecture guide documents an admin-settable pool fee. Current rates and any additional routing-sensitive behavior require deployment-specific reads. No comparable aggregator/direct empirical spread sample was found; public configuration capability does not establish actual charged rates.

### Known weaknesses

Privileged oracle/configuration roles and external-reference validity are material dependencies. Engineering inference: broader assets require per-asset price validation and emergency inventory limits. The reported zero volume may reflect activity or indexing; its cause is UNVERIFIED and cannot support a product-quality ranking.

## Historical venues and universe boundary

The Blockworks overview, retrieved 2026-10-26, flags Obric, Lifinity, Aatrox, Whalestreet, Obsidian, Rubicon and Kipseli inactive, alongside the contradictory Aquifer flag already discussed. DefiLlama's Solana overview explicitly returns zero in its current SolFi V1 and Obric volume fields. These source labels do not prove a program is disabled. Their historical presence is acknowledged; they are not added to current volume as if active. GoonFi's generations are aggregated, while SolFi V1/V2 remain distinct. A permissionless AMM, orderbook or router is not included solely because it serves market makers. Archer/Deriverse and similar venues need a separate mechanism classification before adding them to the proprietary cohort.

## What offchain computation and quote bundling actually save

**Engineering recommendation:** keep prediction, reference-feed fusion and hedge optimization offchain. Publish only the authenticated price/risk state that settlement needs. Store depth shape separately so a fair-value move can reprice standing liquidity without rewriting each level. Verify bounds, expiry, sequence and permissions onchain. Reducing arithmetic at settlement is useful only if oracle validation and minimum-output enforcement remain correct.

Three separate mechanisms must not be conflated:

| Mechanism | Evidence | Benefit to test | Main tradeoff |
|---|---|---|---|
| Shared oracle event reprices pools | HumidiFi homepage litepaper, retrieved 2026-10-26 | Amortized update transaction/signature and shared feed work | Correlated stale state; shared writable accounts can contend |
| Multiple markets packed into one update | Pine's AlphaQ sample; Flint quoting docs, dates below | Lower overhead per refreshed market | Larger account sets/messages; batch delay; failure coupling |
| Builder inserts the latest signed update before a trade | LambdaClass Factory, 2026-10-14 | Avoid publishing every discarded offchain quote | Builder access and ordering dependence; not demonstrated for these Solana venues |

Do not confuse multi-market quote packing with a relay bundle of separate transactions. The latter can help sequencing and inclusion but does not automatically eliminate per-transaction fees or guarantee a lower compute bill. On Solana, measure compute units, requested compute limit, base/priority fee, tips, bytes, writable-account contention and landing success separately. EVM gas is a different unit.

Independent sampled update costs below come from [Pine Analytics' original study](https://pineanalytics.substack.com/p/who-flint-wins-for), published **2026-08-25**, sampling **2026-08-23 12:00–12:10 UTC**. Values are medians per update transaction unless indicated, not timeless benchmarks or whole-transaction dollar costs.

| Provider | Measured CU | Scope/caveat |
|---|---:|---|
| Flint | 281 | Per update transaction |
| BisonFi | 404 | Per update transaction |
| Aquifer | 486 | Per update transaction |
| Quantum | 638 | Per update transaction; cadence is a separate cost |
| AlphaQ | 2068 | Batched across up to 16 markets; roughly 130 CU per market in the stated example |

AlphaQ's per-market figure is amortized, not evidence of a faster single-market update. For comparison, Flint's [current quoting docs](https://docs.flintlabs.dev/sdks/quoting), retrieved **2026-10-26**, describe an unrolled path up to **32** fairs, with a more expensive marginal path beyond that threshold. **Recommendation:** choose batch size using measured latency and cost under the target account set, rather than maximizing the number of markets in each transaction.

## Engineering choices for asset and distribution expansion

| Choice | Build implication | Evidence versus inference | Acceptance measurement |
|---|---|---|---|
| Native majors before arbitrary breadth | Integrate distinct SOL/stable and stable/stable inventory limits and reference checks | The volume baseline supports distribution; profitability remains unknown | Executed depth, net maker markout after hedge/landing costs, stale fills and failed swaps |
| Correlated-asset support | Add separate LST/SOL exchange-rate handling and wrapped-asset basis limits | AlphaQ's observed LST and wrapped-BTC pairs demonstrate coverage; the risk policy is this memo's recommendation | Redemption/basis shocks, reference outages, inventory drawdown and true net exposure |
| Memecoins | Require an observable external reference and feasible hedge/unwind path before quoting | HumidiFi/ZeroFi snapshots establish trading, not safe coverage of all tokens | Price impact under liquidation, unavailable hedges, route concentration and worst loss episodes |
| Tokenized stocks | Model trading-session gaps, reference licensing, redemption/transfer restrictions and distinct issuer mints | Observed labels and AlphaQ's identified SPYx ticker; other backing remains UNVERIFIED | Closed-session behavior, issuer halt/depeg, transfer compatibility and executable unwind |
| Aggregator adapters | Offer reproducible quote simulation with exact CPI context, account metas, token-program compatibility and net recipient checks | Routed flow is observed; the interface requirements are engineering recommendations | Quote-to-fill error, inclusion/revert rate, net user output and router-specific adverse selection |
| Wallet distribution | Test wallet→router→venue end to end and expose clear net output | Dashboard platform attribution is indirect, not proof of bilateral integration | Successful routed fills and retained profitable flow, separated from wallet fees |
| New chains | Separate pricing/risk service from chain-specific settlement and landing adapters | Ethereum's documented builder insertion differs from Solana quote posting | Same fair-value inputs, matched execution-size buckets, full landing bill and degraded-mode behavior |
| Multi-maker shared venue | Evaluate sharing infrastructure if access/onboarding dominates the cost of a proprietary settlement program | Flint documents this option; scale superiority is unproven | Maker isolation, pro-rata allocation, shared-account contention and outage concentration |

**Build sequence:** instrument quote-versus-fill and maker P&L attribution first; prototype compact fair updates and batched refreshes next; add router adapters and validate indirect wallet paths; expand assets only after reference/hedge and failure-mode tests pass. Use a small controlled risk envelope before expanding inventory. These are proposed implementation steps, not claims that this research tested a trading system.

**Required replay scenarios:** delayed/reordered updates; an older signed update arriving after a newer one; a restarted quoting process; a routed simulation with a different invocation context at fill; stale or divergent reference feeds; concentrated same-asset flow; token transfer extensions; a market halted offchain while its Solana token still transfers; and an unavailable fallback venue. A cheaper update is useful only when these cases fail safely.

## Source register and re-pull instructions

All URLs were consulted or queried on **2026-10-26**. Publication/sample dates are recorded separately above. These instructions describe the measurements; no external scripts or dependencies are required to read the delivered CSVs offline.

| Dataset / source type | Exact source or identifier | Re-pull / interpretation |
|---|---|---|
| DefiLlama daily provider series; independent indexer | Per-row `source_url`, `summary/dexs/{slug}` with `dataType=dailyVolume` | Read `totalDataChartBreakdown[].Solana`, not multi-chain total; retain original timestamps and source zeros |
| Chain denominator; same indexer | [Solana overview API](https://api.llama.fi/overview/dexs/solana?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) | Join `totalDataChart` by UTC timestamp; chain coverage is indexer-defined |
| Tessera adapter; primary source code | [DefiLlama adapter](https://github.com/DefiLlama/dimension-adapters/blob/master/dexs/tessera/index.ts) | Solana SQL is `sum(amount_usd)` from `dex_solana.trades` where `project = 'tessera'` and the requested time range |
| GeckoTerminal pools; primary indexer | Pair CSV exact URLs for HumidiFi/ZeroFi | Read `attributes.volume_usd.h24`; include base/quote tokens; paginate; deduplicate pool address; group by case-sensitive mint pair |
| CoinGecko AlphaQ; primary data publisher | [Tickers API](https://api.coingecko.com/api/v3/exchanges/alphaq/tickers?page=1), [field documentation](https://docs.coingecko.com/reference/exchanges-id-tickers) | Read `converted_volume.usd`, not `volume`; retain `is_anomaly`, `is_stale`, timestamps and coin IDs; API values differ from separately refreshed web-page values |
| Blockworks overview; primary dashboard | [Solana PropAMMs](https://app.blockworks.com/analytics/solana/solana-prop-amms), dashboard `582`, chart `9177`, query `4687` (`solana-propamm-table`) | Rounded table values; exact endpoints unavailable here; do not merge with calendar series |
| Blockworks pair amounts; primary dashboard | [Tessera](https://app.blockworks.com/analytics/tessera), chart `7551`, query `3983` (`tessera-top-markets`) | Filter blockchain=Solana. Capture the execution timestamp before treating snapshot amounts as a dated period |
| Blockworks aggregate daily series | Same Solana dashboard, chart `8658`, query `4477` (`solana-prop-amm-volume-by-dex`) | Exact metadata recovered from public page; result rows were unavailable. Metadata is not fabricated numeric data |
| Blockworks pair queries without usable exports | Table below | Query identifiers allow a later authorized dashboard export; unknown amounts remain empty |
| Academic measurements; authors' research | [arXiv v1](https://arxiv.org/html/2609.38056v1), Figure `2(b)` and Table `1` | Pair-restricted daily-share and spread statistics; figure curves were not digitized or invented |
| Technical primary indexer | [Bitquery RFQ API](https://docs.bitquery.io/docs/blockchain/Solana/solana-rfq-api/) | Instruction-based classification is a heuristic, not decoded algorithms |
| Original empirical update-cost research | [Pine: Who Flint Wins For](https://pineanalytics.substack.com/p/who-flint-wins-for) | Short sample and historical catalogue rates; no public raw query ID exposed in examined page |
| Vendor specifications | HumidiFi homepage, AlphaQ site, Flint docs, WOOFi guides, LambdaClass posts linked above | Claims are labelled; no assertion of independent implementation audit |

| Provider | Blockworks dashboard URL | Pair chart ID | Query ID / slug |
|---|---|---|---|
| BisonFi | https://app.blockworks.com/analytics/bisonfi | 10214 | 5356 / `bisonfi-solana-spot-trading-volume-by-pair` |
| Tessera Solana | https://app.blockworks.com/analytics/tessera | 7549 | 3981 / `tessera-solana-spot-trading-volume-by-pair` |
| GoonFi | https://app.blockworks.com/analytics/goonfi | 10260 | 5380 / `goonfi-solana-spot-trading-volume-by-pair` |
| Scorch | https://app.blockworks.com/analytics/scorch | 10204 | 5351 / `scorch-solana-spot-trading-volume-by-pair` |
| ZeroFi | https://app.blockworks.com/analytics/zerofi | 10250 | 5375 / `zerofi-solana-spot-trading-volume-by-pair-2` |
| AlphaQ | https://app.blockworks.com/analytics/alphaq | 7566 | 3988 / `alphaq-solana-spot-trading-volume-by-pair` |
| Aquifer | https://app.blockworks.com/analytics/aquifer | 10881 | 5847 / `aquifer-solana-spot-trading-volume-by-pair` |
| Quantum | https://app.blockworks.com/analytics/quantum | 10270 | 5385 / `quantum-solana-spot-trading-volume-by-pair` |

**What is still unanswered:** current executable pair depth for most providers; exact signed/state quote formats; oracle and pause/withdrawal authorities; router admission terms; current wallet connections; classifier spoof resistance; complete inventory/hedging cost; and deployment-specific audits. Public volume alone cannot answer those. This package deliberately stops short of declaring any provider the best pricing model or safest venue.

## Comparison across providers

The following table is the aligned engineering comparison. Daily USD is for **2026-10-24**; calendar-sum USD and chain share use **2026-09-25/2026-10-24**. All amounts and percentages in those columns are derived from the row's linked DefiLlama API and the dated chain denominator, retrieved **2026-10-26**. P/O refers to the qualified evidence classes above. Technical and integration evidence is linked in each provider section. “Unknown” means UNVERIFIED, not absence.

| Provider | Pricing/update distinction | Asset evidence of engineering interest | Observed routing paths | Daily USD | Calendar-sum USD | Chain share % | Source / as of | Main unresolved issue |
|---|---|---|---|---:|---:|---:|---|---|
| BisonFi | P; private pool state | SOL/stables; wrapped majors | Jupiter/DFlow/Titan/OKX | 219280910 | 8422622566 | 13.281073 | [API](https://api.llama.fi/summary/dexs/bisonfi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Exact pricing/update semantics |
| Tessera V | P; Solana mechanics undisclosed | SOL/stables; wrapped majors; memes | Jupiter/DFlow/Titan/OKX | 135526658 | 3598293558 | 5.673910 | [API](https://api.llama.fi/summary/dexs/tessera-v?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Avoid borrowing Base implementation facts |
| Scorch | P; separate price program | SOL/stables; wrapped BTC; memes | Jupiter/DFlow/Titan/OKX | 60974835 | 2411918532 | 3.803194 | [API](https://api.llama.fi/summary/dexs/scorch?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Quote-state decoding and pair export |
| GoonFi | O heuristic; generations aggregated | SOL/USDC confirmed | Jupiter/Titan/OKX | 45559492 | 2179098168 | 3.436075 | [API](https://api.llama.fi/summary/dexs/goonfi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Current rules and incident remediation |
| HumidiFi | P; shared oracle tick (vendor) | Wrapped majors; memes; stock label | Vendor lists Jupiter/DFlow/Titan/OKX | 39521743 | 1066319584 | 1.681408 | [API](https://api.llama.fi/summary/dexs/humidifi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Model and live fee/classifier rules |
| SolFi V2 | O heuristic; version-specific program | SOL/USDC; other version scope unresolved | Provider-level Jupiter/DFlow/Titan/OKX | 14447600 | 981187417 | 1.547169 | [API](https://api.llama.fi/summary/dexs/solfi-v2?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Version-level integrations and assets |
| AlphaQ | Oracle-driven vendor design; batching observed | LST/SOL; stablecoins; wrapped BTC; SPYx | Jupiter/DFlow/Titan/OKX | 21149410 | 957932388 | 1.510500 | [API](https://api.llama.fi/summary/dexs/alphaq?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Batch semantics and anomalous ticker |
| ZeroFi | O heuristic; multiple pools per pair | Wrapped majors; memes; stock labels | Jupiter/DFlow/Titan/OKX | 28610010 | 764304463 | 1.205181 | [API](https://api.llama.fi/summary/dexs/zerofi?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Uneven indexer coverage |
| QuantumAMM | P; publishing cadence observable | SOL/USDC; rest unknown | Jupiter/Titan/OKX | 4146420 | 707297793 | 1.115291 | [API](https://api.llama.fi/summary/dexs/quantumamm?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Feed validity and full landing bill |
| Aquifer | P; specification unavailable | SOL/USDC; rest unknown | Historical router presence only | 11014298 | 444089280 | 0.700255 | [API](https://api.llama.fi/summary/dexs/aquifer?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=false&dataType=dailyVolume) / 2026-10-26 | Conflicting active/inactive labels |

The smaller venues below have different snapshot windows, so their amounts are intentionally excluded from the aligned numerical columns above.

| Provider | Distinguishing choice | Quantitative status / source date | Main unresolved issue |
|---|---|---|---|
| Riptide | Equity-labelled top pair; model unknown | Rounded dashboard observations in CSV, retrieved 2026-10-26 | Program/operator join and actual stock-market mechanics |
| Taurusfi | SPCX focus in snapshot; model unknown | Rounded dashboard observations in CSV, retrieved 2026-10-26 | Instrument identity and present router reach |
| Flint | Multi-maker, pluggable quotes and pro-rata allocation | Rounded dashboard observations in CSV, retrieved 2026-10-26 | Shared-execution reliability and live catalogue |
| WOOFi sPMM | Per-token pool, custom oracle with Pyth validation | Explicit indexer zero observations in CSV, retrieved 2026-10-26; activity unresolved | Live Solana coverage and comparable execution measurements |

## Design choices that differ

- **Private single-maker engine versus shared multi-maker execution:** most observed venues keep the model private; Flint separates maker strategy from common matching/landing infrastructure. Choose who owns operational complexity and whether shared execution is acceptable.
- **Synthetic impact versus explicit depth ladder:** WOOFi publishes a general notional-sensitive sPMM model; Flint exposes offset ladders and explicit orders. Both can separate a fast fair-value refresh from slow shape/risk changes. Other Solana venues' undisclosed curves cannot be ranked on documentation alone.
- **Shared anchor versus full book refresh:** HumidiFi documents shared oracle repricing; Flint supports persistent offset ladders as well as explicit orders. Store slow-changing depth separately from fast-changing fair value.
- **Update transaction versus execution-time reference access:** indexer heuristics distinguish these, but AlphaQ shows the boundary is not exclusive. Trace the real dependency graph before optimizing.
- **Separate oracle program versus combined venue attribution:** Scorch makes the indexing/monitoring distinction explicit. Measure every update path the swap actually consumes.
- **Uniform advertised fee versus risk-sensitive executable price:** historical routing outcomes vary; current discrimination rules are usually undisclosed. Make quote simulation, effective fees and net settlement consistent.
- **Native continuous markets versus basis/session-sensitive assets:** AlphaQ's LST and wrapped-asset activity and the observed stock labels require different risk controls. A new token or chain should reuse the infrastructure while getting its own reference, inventory, halt and fallback policy.