See the trade before replay
TxStream first against Yellowstone gRPC in matched races, one benchmark run
Median lead over Yellowstone gRPC when TxStream arrives first
Transaction-status accuracy of real-time simulation
Max transactions per message on the batch stream, sent without waiting to fill
One request format,
three ways to take it
Decoded transactions, filtered on our side
RPC Fast reconstructs, recovers, deshreds, decodes and filters in one path, before replay metadata is available. You filter by vote status, signature and included, excluded or required accounts before data crosses the network. Lookup-table addresses arrive when resolved, and the alt_resolution field marks full or partial resolution.
TxStream docs
Vote, signature and account filters
Lookup tables when available, full or partial
signatures_only for timing workloads
Real-time simulation, opt-in
Add simulation_config and each matching transaction comes with a simulation result: predicted status, plus an error when it fails. If capacity or bank state runs short, the status says no outcome is available. Switch on the details you need: compute units, logs, account and token balance deltas, CPI instructions, return data. Status matches the actual execution in approximately 95% of cases. Streams without simulation stay on the immediate path.
Simulation docs
~95% on transaction status
Pick only the details you need
791 µs median cost, full payload
Batches for high throughput
The batch stream sends up to 64 matching transactions per message as soon as they are available, without waiting for the batch to fill. Fewer protobuf and gRPC messages for high-volume consumers; the single stream stays the pick for the lowest per-transaction delay.
Batch docs
SubscribeTransactionBatches
Order kept within and across batches
Same filters and simulation
Who arrives first, with the method
TxStream vs Yellowstone gRPC
First on 99.8%
Median lead 5.02 ms, 17.08 ms at p95. 103,073 shared transactions.
TxStream vs ShredStream gRPC
Neck and neck
ShredStream first on 50.9%, TxStream on 49.1% of 58,688 shared transactions. ShredStream's median lead: 30 µs.
Cost of simulation
+791 µs median
Median
p90
Full payload
791 µs
1.62 ms
Signatures only
832 µs
1.72 ms
Matched-signature comparison: simulation on against the same stream without it.
Coverage in this run: TxStream 99.9%, Yellowstone gRPC 100.0%, ShredStream 56.9% because the benchmark collector does not resolve lookup tables. Coverage describes this run and filter, not a general availability guarantee.
Who TxStream is built for
HFT bot developers, trading platforms, wallets tracking pending transactions, analytics and copy-trading platforms. No shreds to decode, no transactions to reconstruct.
MEV searchers, arbitrage and liquidation bots, market makers and risk engines. Fewer wasted transactions and false positives before block inclusion.
TxStream next to the other feeds
Need confirmed metadata, account updates or blocks? That is Yellowstone gRPC.
TxStream is on the Aperture plan
Support times are the first answer during support chat hours; outside them requests go best effort.
What TxStream does not do
Worth knowing before you plan the integration.
Aperture TxStream FAQ
TxStream vs Yellowstone gRPC: what is the difference?
TxStream is built from validator shreds and sends decoded transactions before replay metadata is available. Yellowstone gRPC comes from the Geyser interface after validator processing and carries confirmed execution metadata, account updates and blocks. Use TxStream for arrival time, Yellowstone gRPC for confirmed state.
Can I simulate Solana transactions inside a gRPC stream?
Yes, as an option: simulation is off until you add simulation_config to the request and pick the details you need: compute units, logs, account and token deltas, inner instructions, return data. Transaction-status accuracy is approximately 95% compared with actual execution results.
Can I get Solana transactions in batches over gRPC?
Yes. SubscribeTransactionBatches sends up to 64 matching transactions per batch, without waiting for the batch to fill, so you get fewer gRPC messages. For the lowest per-transaction delay, use SubscribeTransactions. Filters, signatures_only and simulation work in both.
How do I connect to the TxStream gRPC endpoint?
Point the official Rust client, dysnix/aperture-grpc-client, at aperture-txstream.rpcfast.com:443 with your token in X-Token. It handles keepalives and reconnection; simulation needs client 0.6.1 or later. For a quick test, grpcurl with the official txstream.proto works too.
How much does Aperture TxStream cost?
$499 a month on the Aperture plan, with up to 25 TxStream streams, 10 ShredStream gRPC streams and 25 Yellowstone gRPC streams. The shared infrastructure runs in Frankfurt, Equinix FR13.
