Pinned benchmark release
Solana trading infrastructure
Pump.fun + PumpSwap closed betaLow-latency infrastructure for Solana trading products.
SolTrench streams supported Pump.fun market events, prepares supported Pump.fun or PumpSwap buys and sells, and tracks what happens next. Your wallet signs and your Solana provider sends—your private key never reaches us. SolTrench keeps its own measured work fast; signing, routing, and landing remain under your control.
Free closed beta · customer payments disabled. This invite-only beta is unavailable to anyone currently located in or ordinarily resident in the United Kingdom.
- Your keys stay with you
- You choose how trades are sent
- Pump.fun data · PumpSwap builds
Built for trading products
The hard low latency infrastructure, ready for your product.
Start with Pump.fun market data and Pump.fun or PumpSwap transaction builds today. SolTrench prepares the supported transaction and follows the result while your product keeps control of every decision and signature.
See launches and trades live
Give your app one versioned stream for new tokens, trades, wallet activity, and transaction updates.
- New Pump.fun launches
- Token and wallet activity
- Consistent data your app can use
Get transactions ready to sign
Ask for a buy or sell and receive a prepared transaction that only your customer’s wallet can sign.
- Pump.fun and PumpSwap buys and sells
- Standard and Fast modes
- Premium Fast Entry Protection for Pump.fun buys
- No private key ever shared
Know what happened next
Follow the signature your customer submits. Pump.fun can report the first matching trade event; PumpSwap reports its first-seen status through RPC tracking before the final outcome.
- Pump.fun first matching trade event
- PumpSwap RPC first-seen status
- Confirmed or failed outcome
No private keys cross this boundary.
Build on SolTrench
Build the product. Keep control of the transaction.
Request an invite to the Pump.fun and PumpSwap closed beta, then connect a wallet and keep signing and submission inside your own stack.
Developer onboarding
Wallet-linked API keys with scoped, metered access.
Prove wallet ownership with a signed message, then create an API key for transaction builds, data streams, realtime confirmation, and approved fast mode. During closed beta, include your invite code on first verification. Each key builds trades only for the wallet it is linked to.
Step 1
Verify wallet ownership.
Origin-verified sign-in: supported wallets use Sign In With Solana so they can verify
this page's exact domain before you approve it. The legacy fallback signs the same standardized request
shown below. This signature only proves wallet ownership. It does not create a transaction, grant token
approval, or move funds. Check that the address bar shows exactly soltrench.io; background
API calls use api.soltrench.io, which does not serve the wallet interface. SolTrench will
never ask for a seed phrase or private key. Creating the request may ask your selected wallet to connect;
signing happens only after you press Sign In.
Detecting compatible Solana wallets…
No compatible browser wallet found. Install or unlock a Wallet Standard-compatible Solana wallet, such as Solflare, Phantom, Backpack, or Trust Wallet; then refresh this page. Manual and hardware signing remain available.
Required. Your residence is not inferred from your network location; enter it truthfully and review each declaration before creating the signed message. See how SolTrench uses this information in the Privacy Notice.
No wallet verified yet.
Step 2
Create a scoped API key.
Create or select a wallet-linked API key to use Trade Lab.
Standard and Fast are the supported unsigned-build modes for approved closed-beta accounts. Release qualification is verified during each deployment and is not a promise about customer signing, RPC submission, or landing. Standard keys are self-service; Fast and enterprise access is enabled by an operator for approved accounts.
New Key
{}
Existing Keys
[]
Closed-beta access
Review usage entitlement.
Self-service payments are disabled. Current credit balances are operator-managed access entitlements, not consumable usage units. SolTrench is not accepting customer funds from this page while metering and settlement semantics are being finalized.
Usage entitlement
{"status":"operator_managed"}
Local Transaction API · Solana
Build unsigned Pump.fun and PumpSwap trades for local signing.
Build unsigned buy and sell transactions from a wallet public key. SolTrench never
receives signing material, never signs, and never broadcasts on your behalf. Set confirm=true
to create an intent you can track through WS /v1/data.
Current v1 scope covers Pump.fun v2 bonding curves quoted in native SOL or canonical USDC, plus exact-in
PumpSwap buys and sells for WSOL-quoted pools. Both use this route and the existing
standard or fast mode; select PumpSwap with
pool: "pumpswap" and the exact poolAddress. USDC-quoted PumpSwap pools and
additional venues are not supported yet.
SolTrench is independent and is not affiliated with, endorsed by, or operated by Pump.fun or any other venue referenced in compatibility documentation.
POST /v1/trade-local
X-API-Key: st_live_...
Pump.fun and PumpSwap share one closed-beta transaction API. PumpSwap does not add separate Standard or Fast endpoints, and the same local-sign custody boundary applies to both venues.
SolTrench Entry Protection
Buy now—or not at all.
Speed tells you how quickly a transaction is built. Entry Protection controls how long its unsigned transaction is allowed to remain eligible. Set a validity budget for a Pump.fun Fast buy; SolTrench returns a constrained unsigned transaction or fails closed instead of silently stretching the window.
Interactive testing
Want to try the flow before integrating it?
Open Trade Lab to preview a current Pump curve or PumpSwap pool, build an unsigned transaction, and inspect realtime tracking in one guided workspace.
Request Body
Pump.fun retains its SOL, canonical-USDC, raw-unit, and percentage inputs. PumpSwap uses exact-in raw units only.
{
"publicKey": "Your wallet public key",
"action": "buy",
"mint": "Token mint address",
"amount": 0.0001,
"denominatedInSol": true,
"mode": "standard",
"slippage": 8,
"priorityFee": 0.001,
"pool": "pump",
"confirm": true,
"computeUnitLimit": 400000,
"closeAccount": false
}
Canonical USDC buy (1.50 USDC in six-decimal raw units):
{
"publicKey": "Your wallet public key",
"action": "buy",
"mint": "USDC-quoted Pump token mint",
"quoteMint": "EPjFWdd5AufqSSqeM2qN1xzybapC8G4wEGGkZwyTDt1v",
"quoteAmountRaw": "1500000",
"denominatedInSol": false,
"mode": "standard",
"slippageBps": 800,
"pool": "pump"
}
PumpSwap exact-in WSOL buy through the same transaction route:
{
"publicKey": "Your wallet public key",
"action": "buy",
"mint": "PumpSwap base-token mint",
"pool": "pumpswap",
"poolAddress": "Exact PumpSwap pool address",
"quoteAmountRaw": "1000000",
"denominatedInSol": true,
"mode": "standard",
"slippageBps": 800
}
For a PumpSwap sell, replace quoteAmountRaw with tokenAmountRaw, set
action: "sell", and set denominatedInSol: false. Percentage sells and Pump-only
Entry Protection or recovery fields are not accepted. Historical Pump.fun benchmark cohorts do not apply to PumpSwap.
Fast buy with premium Entry Protection:
{
"publicKey": "Your wallet public key",
"action": "buy",
"mint": "Token mint address",
"amount": 0.0001,
"denominatedInSol": true,
"mode": "fast",
"slippageBps": 800,
"confirm": true,
"entryProtection": {
"enabled": true,
"validityBudgetMs": 1200
}
}
Full-exit sell with an explicit, caller-directed recovery workflow:
{
"publicKey": "Your wallet public key",
"action": "sell",
"mint": "Token mint address",
"amount": "100%",
"denominatedInSol": false,
"mode": "fast",
"confirm": true,
"closeAccount": true,
"sellCurveMode": "fresh",
"allowRpcFallback": true,
"sellSlippageLadder": {
"enabled": true,
"step": 1
}
}
For a permitted successor step, add the immediately preceding attempt proof:
{
"sellSlippageLadder": {
"enabled": true,
"step": 2,
"previousAttempt": {
"intentId": "st_int_...",
"signature": "LocallyDerivedSolanaSignature..."
}
}
}
buy or sell.standard for ordinary builds, fast for eligible approved keys.100% for sells.amount.pump selects the Pump.fun bonding curve; pumpswap selects PumpSwap on this same route.pool: "pumpswap"; use the exact PumpSwap pool public key.slippageBps for exact basis points.enabled and an estimated upper validityBudgetMs. The response returns authoritative cutoff evidence, and the request fails closed when the budget cannot be safely met. This bounds transaction validity; it does not guarantee signing, landing, or execution.stepsBps only when supplying your own ordered basis-point limits; otherwise the service policy applies.intentId and locally derived signature. A successor is accepted only after SolTrench verifies that the preceding attempt is no longer eligible to land./v1/data.Examples
Build, sign, arm, broadcast.
Trade Lab · venue-aware closed beta
Preview a Pump.fun or PumpSwap trade before you build it.
Start with a read-only point-in-time quote, then choose whether to build an unsigned transaction. Preview creates nothing; Build never signs or broadcasts; tracking starts only after you explicitly arm it.
POST /v1/pump/quote
POST /v1/pumpswap/quote
POST /v1/trade-local
WS /v1/data
Market preview
Read current Pump.fun curve or PumpSwap pool state, fees, slippage bounds, freshness, and buildability without receiving transaction bytes or creating a confirmation intent.
Unsigned build and tracking
Build for local signing, then arm realtime tracking with the actual signature you derive and broadcast.
Build from a working example
Open the Examples tab for complete Python projects covering live data, read-only quotes, unsigned builds, and locally signed flows.
Read-only quote API
Know the current sizing before building.
Choose POST /v1/pump/quote for Pump.fun bonding curves or
POST /v1/pumpswap/quote for an exact PumpSwap pool. Both return decimal-string raw amounts
and return no transaction bytes or intent, signature, or broadcast.
Pump.fun quote request:
{
"action": "buy",
"mint": "Token mint address",
"quoteAmountRaw": "100000000",
"slippageBps": 800
}
PumpSwap exact-in WSOL buy quote request:
{
"publicKey": "Wallet linked to this API key",
"pool": "Exact PumpSwap pool address",
"action": "buy",
"quoteAmountRaw": "1000000",
"slippageBps": 800
}
Both requests use raw exact-in units. Sells replace quoteAmountRaw with raw
tokenAmountRaw. A preview is not a price lock: the later build independently uses then-current state.
Interactive API workbench
Test one trade setup, one deliberate step at a time.
Pump.fun quotes need the market inputs only. PumpSwap quotes also need the pool address and wallet linked to your API key. Both venues use the same Standard/Fast selector and unsigned build action.
- 1PreviewRead-only current sizing
- 2BuildUnsigned transaction only
- 3TrackAfter your local signature
Track a transaction you signed locally
Paste the actual signature only after local signing. Arm tracking before you broadcast through your RPC.
Build returns an unsigned transaction. Sign locally, derive the real signature, arm tracking, then broadcast through your own RPC.
Benchmark context
Choose standard or fast.Unsigned build API response
{}
Live activity
Optional creation events and status messages for this browser tab.
Pump.fun Paper Lab · scenario stress desk
Pick a path. Stress the exit. See exactly what the model did.
Choose built-in synthetic trade scenarios or bring your own trajectories. Apply fast exit conditions and inspect every selected observation without placing an order.
- 1Choose trade pathsBuilt-in scenarios or your pack
- 2Set the conditionsSelect paths, timer, bracket, or custom
- 3Inspect the outcomeChart, ledger, and evidence limits
Step 1 · choose your source
Start with built-in scenarios or bring your own.
The built-in library is the fastest route into the stress desk.
Open the scenario library
Load five invented 250 ms paths: spike, climb, whipsaw, bleed, and a deliberately short tape.
- Select one or several
- Works with every exit recipe
- Invented learning data
Bring your own paths
Use compact trajectory JSON, then choose which candidates receive the paper conditions.
- Up to 8 MB
- Uncompressed UTF-8 JSON only
- Nothing is uploaded
Use prepared Pump.fun history
Preview the replay-ready packs we plan to offer after provenance and rights review.
- Compact trajectory shards
- Coverage manifest
- Browser-ready format
Planned 24-hour Pump.fun-compatible Replay Pack
Planned as compact per-launch event-price trajectories with exact UTC coverage, time-source labels, gap counts, schema notes, field-removal notes, and SHA-256 checksums.
The saved corpus covers creations from 20 Jun 2026 15:37:58 UTC to 21 Jun 2026 10:42:13 UTC and 1,432,564 tracked events through 22 Jun 2026 08:05:08 UTC. It is not a complete 24-hour all-market feed and cannot be offered until completeness, transformation, provenance, privacy, and redistribution-rights checks are complete.
- Planned window
- 24 UTC hours, not yet qualified
- Planned delivery
- Compact trajectory shards
- Price
- Not set
Independent product preview; not affiliated with or endorsed by Pump.fun.
Checkout is disabled and this page does not take payment. Request access to hear when a qualified pack is available.
Request early accessYour launch data · optional
Upload or paste a trajectory pack.
Required per launch: creation_event_t0_ms, timestamp_status: MS_SAFE, and ordered
events with available_at_ms, sequence, and price. The _ms
values must be integer epoch milliseconds; seconds are rejected. A user-supplied MS_SAFE
label is not SolTrench certification of provenance, coverage, or clock quality.
Only uncompressed UTF-8 JSON is opened. Archives, binary files, excessive nesting, and oversized packs are rejected without being unpacked. A pack can contain up to 500 paths, 5,000 observations per path, and 50,000 observations overall. SolTrench does not upload or persist the file; validated records stay in this tab and the raw text is discarded after loading.
Choose the sample or add a compatible Pump.fun pack.
Step 2 · stress the paths
Choose the paths, then set the exit conditions.
Select one or several loaded paths. The chart previews the active path before a run and adds the exact observations selected by the recipe afterward.
Choose what to stress
Each candidate is a supplied price-observation path, not an order. Select multiple paths to run them in timestamp order.
Load the built-in scenarios or import a trajectory pack to choose paths.
No paths loaded.
Trajectory stress chart
Load a path to preview its observations.
Choose a recipe
Keep the defaults or open the advanced controls to change timing, thresholds, capital, and friction.
Recipe labels describe modeled mechanics only. They are not safety ratings, recommendations, or predictions.
Selects a later observation after three seconds. Visible timing assumptions: 250 ms decision, 500 ms to the selected entry observation, and 500 ms to the selected exit observation.
Customize this recipe Optional · advanced assumptions
Step 3 · price-path hypothesis
See every observation the conditions selected.
No replay run yet
Run a fast-exit recipe to get a plain-language result.
See all modeled metrics
Warnings
Reasons not to confuse this with live trading
- Run a diagnostic to see timing, cost, and open-exposure limits.
Evidence card
Fast replay provenance
- Source
- Not loaded
- Model
- EVENT_PRICE_HYPOTHESIS · promotion_eligible=false
- Time basis
- Integer epoch-ms availability observations required; user labels are not certified
- Entry
- First selected price observation after decision + entry-observation delay; not an order or execution
- Exit
- Trigger selects the first price observation after the exit-observation delay; not a submitted or landed sell
- Capital
- One hypothetical position at a time; later launch candidates wait until the prior path resolves
- Costs
- Configured aggregate friction only; reserves, live quotes, retries, fees, tips, rent, failed exits, and no-fills are not modeled
Audit trail
Price-path hypothesis ledger
No price paths calculated
| # | Launch | Selected entry observation | Selected exit observation | Observed hold | Reason | Entry price | Exit price | Configured costs | Hypothesis P&L | State |
|---|---|---|---|---|---|---|---|---|---|---|
| Run a fast-exit price path to populate the ledger. | ||||||||||
PumpSwap Paper Lab · completed-candle replay
Test recovery, breakout, and momentum ideas on PumpSwap candles.
Use the existing guided candle lab for established-pool strategies. These longer-horizon recipes are separate from Pump.fun launch sniping and never share venue assumptions or data.
- 1Choose dataDemo, your file, or a data pack
- 2Choose strategyOne click or customize
- 3Read the resultOutcome plus honest caveats
Step 1 · choose data
How would you like to start?
No data expertise required. Pick the easiest option for you.
Try the free walkthrough
Loads an invented candle example so you can learn the page in under a minute.
- No file needed
- Works with every preset
- Learning only—not market data
Upload your own data
Use a JSON or CSV file with one continuous series of explicitly completed OHLC candles.
- Up to 8 MB / 25,000 bars
- Uncompressed UTF-8 only
- Nothing is uploaded
Use prepared PumpSwap candles
We plan to prepare venue-labelled candle packs for these longer-horizon strategies.
- Normalized candle shards
- Coverage and checksum manifest
- Designed to load into Paper Lab
Planned PumpSwap Candle Replay Pack
Planned as a UTC-bounded set of venue-labelled completed OHLC candles with JSON/CSV shards, schema notes, interval and quote-unit labels, gap counts, and SHA-256 checksums.
Coverage, transformation, provenance, privacy, and redistribution rights must be cleared before a pack can be offered.
- Planned window
- 24 UTC hours—not yet collected or qualified
- Use
- Historical Paper Lab replay
- Planned delivery
- Prepared market selector
- Price
- Not set
Independent product preview; not affiliated with or endorsed by PumpSwap or Pump.fun.
No PumpSwap candle pack has been qualified for customer delivery. Checkout is disabled and this page does not take payment.
Request early accessYour data · optional upload
Upload or paste completed candles.
Paste a JSON array or CSV with time, open, high, low,
close, and an explicit completed flag or status. Newest forming bars are excluded.
Keep one market, quote unit, and candle interval per run. Only uncompressed UTF-8 JSON or CSV is opened; archives, binary files, excessive nesting, and oversized structures are rejected without being unpacked. SolTrench does not upload or persist the file. Validated records stay in this tab and raw text is discarded.
Add completed candles or choose the illustrative walkthrough to begin.
Step 2 · choose a strategy
Pick how you want the model to trade.
Choose a ready-made recipe below. We fill in the assumptions for you, and you can fine-tune them later if you want. Every recipe enters only after a completed signal candle, using the next bar's open.
Recipe tags describe only the modeled rules. They are not safety ratings, recommendations, or predictions.
Waits for a sharp fall to stop making lows, then enters after price starts recovering.
Customize this strategy Optional · advanced assumptions
Step 3 · modeled outcome
Read the result with its caveats attached.
No replay run yet
Run a strategy to get a plain-language summary here.
See all modeled metrics
Completed prices, modeled entries, and exits
Balance after modeled closed positions
Warnings
Reasons to distrust the headline.
- Run a diagnostic to see data, ambiguity, capital, and unresolved-exposure warnings.
Evidence card
Replay provenance
- Source
- Not loaded
- Model
- CANDLE_MODELED
- Bars used
- 0 completed
- Bars excluded
- 0
- Signal timing
- Entry at next completed bar open
- Same-bar conflict
- Protective stop first
- Cost boundary
- Configured aggregate friction only; fees, tips, rent, price impact, partial fills, failed exits, and no-fills are not independently modeled
- Open exposure
- Unresolved positions are shown separately and excluded from closed-trade P&L, return, drawdown, balance, and friction
Audit trail
Modeled trade ledger
No modeled trades
| # | Signal | Entry | Exit | Reason | Entry price | Exit price | Costs | Net P&L | State |
|---|---|---|---|---|---|---|---|---|---|
| Run a diagnostic to populate the ledger. | |||||||||
Session comparison
Compare assumptions, not just returns.
Pin selected runs for side-by-side review in this tab. A comparison is not saved to an account and is not evidence that any strategy will execute or remain profitable.
No runs added during this browser session.
| Run | Source | Strategy / SL / TP | Closed | Win rate | Modeled net P&L | Modeled return | Modeled max drawdown |
|---|---|---|---|---|---|---|---|
| Add a completed run to compare it here. | |||||||
Starter projects · Pump.fun data + PumpSwap builds
Start with working code, not a blank file.
Copy a small Python foundation, connect your own API key, and build from there. The current starter library focuses on Pump.fun data and trades; the PumpSwap quick start below uses the same API key and shared unsigned-build route.
15 copyable starters
Every project has complete source
Python · downloadable source
Collect launches
Learn the websocket with no wallet, signing, or transaction flow. Collection begins when you connect.
Your machine signs and sends
Trading examples load a local keypair, arm SolTrench tracking, then broadcast through your own RPC. Your API key must be linked to that same wallet.
Pump.fun data · PumpSwap builds
Live-event starters use the Pump.fun feed. PumpSwap is callable for read-only quotes and exact-in WSOL-quoted unsigned builds; no PumpSwap event stream, historical backfill, or shared paper-account support is claimed.
PumpSwap quick start
Preview the pool, then build on the existing route.
Use POST /v1/pumpswap/quote with the exact pool, wallet, action, slippage and raw exact-in
amount. To build, send the wallet, action, base-token mint, exact-in amount and slippage to
POST /v1/trade-local with pool: "pumpswap", the exact
poolAddress, and your existing standard or fast mode. SolTrench
returns unsigned bytes only.
Copy, download, extend
Fifteen foundations you can run today.
Pick a project to inspect its complete source. Each file uses environment variables for credentials and keeps private signing material outside SolTrench.
Launch buy + 3-second exit
Detect the first live launch, build and sign locally, wait for the buy to confirm, hold three seconds, then repeat the same tracked flow for a full sell.
Loading starter source…
Blueprints upgraded
All fifteen projects now include complete code.
The nine former blueprints are selectable above with the original six. Choose the boundary that matches what you want to learn before adding more automation.
Start without signing
Collectors, monitors, screeners, alerts, candles, and local strategy decisions use live data or read-only quotes.
9 examples · no transaction submissionBuild without private keys
Trade-ticket and scheduled-buyer starters can stop after the preview or save unsigned bytes for your own signing flow.
2 examples · explicit unsigned-build gateOpt into live-funds flows
Round-trip, exit, and latency examples sign locally, arm tracking before broadcast, and require an explicit live flag.
3 live examples · 1 tracking-only watchdogPump.fun Data API
Pump.fun market events plus transaction status over one websocket.
Connect once, subscribe by method, and keep a single socket open. Pump.fun subscriptions carry creation, token-trade, and account-trade events. Transaction intents created from supported Pump.fun or PumpSwap builds can deliver realtime status for the actual locally signed transaction you submit.
PumpSwap has no market-data stream in this release. Its transaction tracking is RPC-based first-seen and final status, not a PumpSwap trade event or migrated-pool data subscription.
POST /v1/data/token
WS /v1/data?token=st_ws_...
subscribeNewToken
trackTransaction
Pump.fun is the current market-data API. Use its event subscriptions for live creates and
trades. trackTransaction also supports a PumpSwap build intent through RPC status tracking.
Historical lifecycle runs remain engineering evidence only. SolTrench does not publish them here as current performance because they were not an exact-release customer cohort. The Benchmarks page identifies the release, sample size, measurement boundary, and limitations for every customer-facing result.
Stream Methods
Mint a short-lived websocket token with X-API-Key, then use one websocket connection for creation events, trade events, and transaction tracking.
standard is the only source accepted by the public v1 websocket. Internal upstream source topology is not customer-selectable or part of this API release.
creation_event_t0 and public token metadata.
intentId and the signature. Pump.fun may match a trade event; PumpSwap uses RPC status only.Examples
Subscribe and track on one socket.
Payloads
Creation, trade, and transaction tracking messages.
Send the control messages below. Emitted data events use a versioned envelope with explicit canonical units; additional adapter fields may be present under data.
{
"method": "subscribeNewToken",
"source": "standard"
}
{
"method": "subscribeTokenTrade",
"source": "standard",
"keys": ["MintAddress111111111111111111111111pump"]
}
{
"method": "trackTransaction",
"intentId": "st_int_...",
"signature": "real_signature_after_local_signing",
"wallet": "Wallet public key",
"mint": "Token mint",
"action": "buy"
}
{
"type": "tokenTrade",
"schemaVersion": "1.0",
"venue": "pump.fun",
"source": "standard",
"eventId": "transaction_signature:0",
"units": {
"solAmount": "SOL",
"solAmountLamports": "lamports",
"tokenAmountRaw": "raw token base units"
},
"data": {
"mint": "Token mint",
"signature": "transaction_signature",
"eventIndex": 0,
"wallet": "Trader wallet",
"side": "buy",
"solAmount": 0.0001,
"solAmountLamports": 100000,
"tokenAmountRaw": 2955913907,
"observedAt": "2026-07-22T10:00:00Z"
}
}
Pinned Pump.fun evidence · exact release
Measured Pump.fun unsigned performance.
Fast where it counts. Clear about what was measured.
Benchmark release qualification status unavailable. This page separates the result, the measurement boundary, and the proof so each number is read in the right context. PumpSwap is callable, but its latency benchmark is still pending and no Pump.fun figure on this page applies to it.
Loading the sanitized evidence record.
Read this page in layers Start with the result. Open the audit detail only when you need it.
The 30-second read
The pinned release passed every measured unsigned request.
Fast mode reduced the median API-owned build step to well below one millisecond. The widest clock measured on the same host—the complete HTTP round trip—had a verified median. These are exact-release Pump.fun beta results, not an SLA or a promise of customer end-to-end latency.
Every measured unsigned build completed.
UnavailableCohort scope unavailable
- Warmup scope unavailable
- Signing scope unavailable
The work SolTrench owns moved first.
Read the HTTP number for the widest measured path. Read the build number only for API-owned transaction construction.
Live launch and trade delivery also qualified.
Qualification unavailableDetector callback or trade decode → same-host benchmark-client WebSocket receipt.
One request · three nested clocks
The larger clock contains the smaller one.
These Fast-mode medians describe progressively wider parts of the same measured request. They are not three competing claims.
- 01 Server build Unavailable Unsigned transaction construction
- 02 Server response Unavailable API ingress → response start
- 03 Same-host HTTP Unavailable Request start → complete JSON response
Same-host benchmark client → AWS us-east-1 (Virginia) production edge → complete unsigned Pump.fun JSON response.
Customer distance, wallet signing, simulation, RPC submission, confirmation, landing, and every PumpSwap path.
p50The middle measured request—the clearest headline for a typical run.
p9595% of measured requests finished at or below this value.
p9999% of measured requests finished at or below this value.
Pinned measured Pump.fun API runtime
Complete p50, p95, and p99 unsigned distributions.
Unsigned p50, p95, and p99—with every denominator visible. Compare Standard and Fast on the same clocks, then open the evidence pack for all eight cohorts, publication checks, and release hashes.
Standard mode
Unavailable measured requests passed
- p50 median
- Unavailable
- p95
- Unavailable
- p99
- Unavailable
- p50 median
- Unavailable
- p95
- Unavailable
- p99
- Unavailable
- p50
- Unavailable
- p95
- Unavailable
- p99
- Unavailable
Fast mode
Unavailable measured requests passed
- p50 median
- Unavailable
- p95
- Unavailable
- p99
- Unavailable
- p50 median
- Unavailable
- p95
- Unavailable
- p99
- Unavailable
- p50
- Unavailable
- p95
- Unavailable
- p99
- Unavailable
Audit-ready detail Open the full evidence pack Scope, 8 cohorts, publication guard, and artifact hashes
Cohort scope unavailable
Excluded from percentile calculations
SOL + USDC · BUY + SELL
SOL + USDC · BUY + SELL
Signing remained customer-side and out of scope
No submission, confirmation, or landing
Client region: Virginia edge. Same-host unsigned HTTP request start → complete response is the widest published same-host edge-path boundary. Server response and server build are separately reported clocks. No signing, simulation, submission, confirmation, or landing is included. Results are retained release-bound beta evidence—not an SLA or global customer latency.
Eight contract-checked cohorts
Every quote asset, side, and mode stays inspectable.
Each row reports measured responses only; warmups are excluded.
| Quote / side | Mode | Succeeded / measured | Same-host HTTP | Server build |
|---|---|---|---|---|
| SOL BUY | Standard | Unavailable | Unavailable | Unavailable |
| SOL BUY | Fast | Unavailable | Unavailable | Unavailable |
| SOL SELL | Standard | Unavailable | Unavailable | Unavailable |
| SOL SELL | Fast | Unavailable | Unavailable | Unavailable |
| USDC BUY | Standard | Unavailable | Unavailable | Unavailable |
| USDC BUY | Fast | Unavailable | Unavailable | Unavailable |
| USDC SELL | Standard | Unavailable | Unavailable | Unavailable |
| USDC SELL | Fast | Unavailable | Unavailable | Unavailable |
- Eligible
- Unavailable
- Release stable
- Unavailable
- Zero measured failures
- Unavailable
- Warmups passed
- Unavailable
- HTTP 429
- Unavailable
- Provider-like 503
- Unavailable
- Post-stop requests
- Unavailable
- Release
Unavailable- Packaged core
Unavailable- Run ID
Unavailable- Completed
- Evidence artifact SHA-256
Unavailable- Canonical payload SHA-256
Unavailable- Fixture SHA-256
Unavailable- Harness SHA-256
Unavailable
Historical funded campaign · separate measurement boundary Outcome counts published. Latency withheld. Historical evidence unavailable
Separate funded lifecycle record
Funded outcome counts remain historical and withheld.
These counts describe every finished funded attempt under the campaign’s predeclared eligibility rules. They are not merged into the pinned benchmark-release unsigned distributions above.
Every finished invocation in the append-only ledger
Eligible under the predeclared rules
One BUY and one SELL per eligible lifecycle
All seven occurred in Fast mode
No retained benchmark position remains open
31 eligible lifecycles per mode
- Release
Unavailable- Packaged core
Unavailable- Campaign
Unavailable- Ledger head
Unavailable- Campaign date
- Verification state
- Not published
Evidence summary unavailable.
Retained funded outcomes
Reliability denominators stay next to every performance claim.
The complete ledger is reported by mode. Success-only latency figures are not published.
| Mode | Finished attempts | Eligible closed lifecycles | Excluded attempts |
|---|---|---|---|
| Standard | Not published | Not published | Not published |
| Fast | Not published | Not published | Not published |
| Mode | Excluded outcome | Count | Position state |
|---|---|---|---|
| Fast | Unavailable | Not published | No funded position retained |
| Fast | Unavailable | Not published | Closed through recovery |
| Fast | Unavailable | Not published | Closed through recovery |
Outcome evidence status unavailable.
Exact measurement definitions
The clock endpoints are part of the result.
These public boundaries describe what was timed without disclosing private implementation topology. A smaller server clock is never presented as a customer’s full round trip.
- Same-host HTTP round trip
- Benchmark-client request start → complete unsigned HTTP response received through the Virginia edge path. This is the widest clock published for the pinned benchmark release.
- Server response
-
API ingress through response start, reported by
X-Response-Time-Ms. It excludes the benchmark client and the remaining edge-path portion of the complete HTTP round trip. - Server build
- Server-recorded unsigned transaction construction only. It is a deliberately narrower clock than server response or same-host HTTP.
- Outside every unsigned clock
- Wallet signing, simulation, submission, confirmation, raw trade-event observation, and landing. Customer network distance and RPC choice are also outside this same-host result.
- Percentile method
-
nearest_rank_v1. Sort ascending; rank =max(1, ceil(percentile × n)), one-indexed. Results are rounded to three decimal places. Warmups are excluded. - Interpretation
- Exact-release beta evidence, not an SLA, not a fastest-run claim, and not a forecast of global customer latency. Each distribution retains its own mode and cohort denominator.
Separate funded boundary: the retained July 25 lifecycle record measured different endpoints after local signing and broadcast. Its latency percentiles remain withheld and are never pooled with this unsigned qualification.
Historical campaign audit Open planned-window coverage W1 complete · W2 partial · W3 not begun
Cohort coverage
Planned-window coverage is not complete.
Coverage is shown against each mode’s predeclared target inside each campaign window.
| Window / mode | Coverage | Progress |
|---|---|---|
| W1 Standard | Not published | |
| W1 Fast | Not published | |
| W2 Standard | Not published | |
| W2 Fast | Not published | |
| W3 Standard | Not published | |
| W3 Fast | Not published |
Historical evidence
Different releases and boundaries stay in different records.
Historical measurements remain available for audit, but never merge into the funded snapshot above.
Unsigned-build benchmark
Exact release b90c5cd3c5bdc124cb830bd7e17260c1a3e03400. Its boundary ends at the
complete unsigned HTTP response and excludes local signing, submission, confirmation, and trade-event
observation.
BUY and SELL raw-landed observations
One successful BUY lifecycle and one successful SELL lifecycle. These are individual observations, not percentiles or a success rate; the release commit is not asserted by the page.
n=1 BUY + n=1 SELL · separate historical boundary.Never merge different release identities or measurement boundaries into one percentile.
Measurement publication roadmap
Published where qualified. Pending everywhere else.
Each pending category remains nonnumeric until its own reviewed, exact-release pipeline exists.
Same-host Virginia edge · exact measured API runtime · 400 measured responses
Published by mode and by all eight unsigned cohorts
Detector callback → same-host WebSocket client · exact-release live launches and trades
Verified evidence is unavailable, so no performance or delivery qualification is shown.
Requires a durable availability series.
Requires qualified customer-path measurements outside the same-host edge boundary.
Requires qualified multi-region evidence.
Requires an auditable reference comparison.
Retained unsigned benchmark: Unavailable. Retained Observe benchmark: Unavailable. Publication states are reviewed before release.
Solana venue coverage
Pump.fun and PumpSwap, one local-sign API.
Pump.fun data, quotes and unsigned builds are customer-callable in closed beta. PumpSwap read-only quotes and exact-in WSOL-quoted unsigned builds are callable through the same account and transaction contract. Additional venues stay behind the public boundary until their contracts and reliability gates are ready.
Current public contract
One local-sign envelope without flattening venue differences.
Check the request and current venue conditions against the published contract.
Return one unsigned transaction with the customer’s explicit amount and protection limits.
Your wallet or secure signer signs locally. Private key material never reaches SolTrench.
You choose the submission infrastructure and retain broadcast control.
Release boundary
Coverage is announced as a customer contract, not an engineering roadmap.
Callable coverage is stated explicitly below. Future market structures remain unannounced until their customer contracts are ready; a listed integration never implies a safe pool or profitable strategy.
Pump.fun bonding curves
Live creation and trade events, read-only quotes, and unsigned SOL- or canonical-USDC-quoted buy and sell builds.
PumpSwap pools
Read-only pool quotes and exact-in WSOL-quoted unsigned buys and sells. Standard and Fast remain modes
on POST /v1/trade-local; PumpSwap data streaming and latency evidence are not published.
Additional Solana venues
Future contracts will publish supported operations and evidence only when a customer-facing adapter is ready.
Evidence boundary
Private venue engineering evidence stays private.
PumpSwap is available to approved closed-beta accounts within the scope above, but its latency benchmark remains pending. Retained Pump.fun cohorts do not apply to PumpSwap. Internal calibration artifacts do not imply a release order, latency performance, or strategy performance.