Solana shred stream providers after Jito ShredStream's shutdown (2026)

Last updated: October 25, 2026

Disclosure: We sell one of these (Triport Shred stream); its row is marked (ours). This is a dated comparison, not a speed ranking.

Shred feed providers: delivery, price and setup

Monthly prices below cover different products. An IP in a data centre is not the same unit as an application connection. Include receiver hosting and deshredding costs when comparing a raw feed.

Raw shreds vs deshredded transactions

Choose raw if your pipeline must inspect shred boundaries or feed packets into a node, including colocated infrastructure. Choose deshredded if you would immediately reconstruct transactions and discard the packets. These are different integration contracts, so changing the endpoint alone is insufficient. For background, read what a shred is and why it arrives first. For the WebSocket schema and plan limits, see Triport Shred stream (deshredded, WebSocket).

What you need for a raw shred feed

• A reachable public IPv4 address without NAT or CGNAT; check the receiver IP before buying an IP allowance.
• An open UDP port, firewall rules for the vendor and a receiver bound to the correct interface.
• Enough sustained bandwidth and packet-processing capacity. Thor specifies 50–100 Mbps for a full feed (Source, read on 2026-10-24); treat this as its published specification, not our measured traffic or a universal ceiling.
• A deshredder, packet verification and FEC recovery on your side, plus handling for incomplete data.
• Provider IP whitelisting, region selection and a procedure for moving or replacing the receiver.

The legacy Jito setup reference, read on 2026-10-24 documents the UDP and NAT model. It is useful for checking infrastructure assumptions, not for obtaining new Jito service. Validate the replacement vendor's requirements before committing a receiver.

Pricing models: per IP, per region, per time interval, included

For one receiver in one region, compare Helius's IP rate, Triton's IP/data-centre rate and Everstake Connect's region/IP allowance in the table. For two regions, price both receivers and both locations explicitly; do not assume an IP allowance includes another data centre. Everstake Ultra bundles region access, but confirm its current IP allowance rather than inferring it from the plan name.

For a day of testing, Triton publishes a price per 15 minutes. Multiply the selected location rate by the billable intervals, then check minimum funding and cancellation terms. This is a costing method, not a verified day-pass offer. Thor includes access with passes; the pass price still matters. ERPC and DoubleZero require current vendor terms before a cost comparison is possible.

For example, at the published monthly entry rates, one-region Everstake Connect ($500) is below Helius Standard ($1,000) and Triton Frankfurt/Amsterdam ($1,500 per IP/data centre). Two Triton receivers, one in Frankfurt and one in Amsterdam, have a nominal monthly sum of $3,000, versus $1,100 for Everstake Ultra's advertised region bundle. One continuously billed day at Triton's Frankfurt/Amsterdam interval rate works out to $56.64 per receiver. These are arithmetic examples from the dated prices above, not equivalent-service quotes; confirm IP allowances, capacity and payment terms.

Triport includes Shred stream in a monthly plan. There is no separate per-GB charge or data-volume quota; connection, account-filter and API RPS limits still apply. Use the table's plan connection counts, not a raw provider's IP count, when budgeting application replicas. Shred stream is absent from Basic and the trial; Enterprise limits are contractual.

What vendors claim about speed

Vendor wording below is attributed to the 2026-10-24 snapshot, not adopted as our result. Timing figures are omitted: this page has no independently measured timing dataset. A claim about a different transport or observation point cannot rank these raw feeds.

How to verify a shred feed yourself

1. Run consumers on the same host, with the same observation window, region and transaction filters. Record client versions, receiver configuration and exclusions before starting.
2. Compare equivalent events. Timestamp a complete transaction when it becomes usable; keep first raw-packet receipt as a separate metric. Otherwise assembly work distorts the comparison.
3. Use one local monotonic clock. Match by signature, deduplicate, and exclude connection warm-up under a rule fixed in advance. Do not subtract a remote server timestamp from your local clock.
4. Report share of first detections, ties, missing matches, packet loss and reconnect gaps. State the sample and denominator, including signatures seen by only one feed.
5. Repeat from the intended deployment region. First detection is not successful execution, inclusion probability or a transaction landing rate.

When a shred feed is the wrong tool

If you need execution statuses, logs or balances, use a commitment-aware channel such as Solana gRPC (Yellowstone). If your consumer needs transactions but you cannot operate a UDP receiver, consider deshredded delivery. In either case, observing intent does not establish a successful trade; confirm the transaction outcome.

When Triport is the wrong choice

Triport does not supply raw UDP packets to a validator, or replay missed Shred stream frames. Slow consumers can lose updates under backpressure. Standard plan regions are EU and US; confirm any custom-location requirement separately. If your integration needs raw shreds, guaranteed replay or another standard region, use those requirements to narrow the provider table first.

Methodology and dates

Competitor prices, units and the Jito sunset record come from the dated provider catalog; supplementary regions, receiver requirements and claims come from the same 2026-10-24 source snapshot. Triport facts were checked against its plan matrix and Shred stream reference on 2026-10-25. “Not verified” means this snapshot cannot establish the feature, not that the vendor lacks it. Refresh prices monthly and verify checkout terms before purchase. No head-to-head measurements were performed.