Pyth Network is an oracle that brings real-world prices onto Solana and other blockchains, sourced directly from over 120 publishers including trading firms like Jane Street, Cumberland and Wintermute alongside venues such as CBOE. It uses a pull model: prices are aggregated continuously on Pythnet, its own appchain, and an application requests the latest update on-chain at the moment it needs one, rather than a feed being pushed constantly whether anyone reads it. That design is why oracle costs scale with usage instead of with time, and it is the main structural difference from earlier oracle networks.
Key Facts
- Over 120 publishers contribute prices, including Jane Street, Cumberland, Wintermute and CBOE.
- Data comes from firms that trade the assets, not from scraping public APIs.
- Pull model: applications request an update when needed instead of receiving constant pushes.
- Aggregation happens on Pythnet, a dedicated appchain, before delivery elsewhere.
- Cross-chain delivery ran over Wormhole; after 31 July 2026 it moves to a 3-of-5 router quorum.
- Each price comes with a confidence interval, not just a single number.
- PYTH traded around $0.049 with roughly $388M market cap in mid-July 2026.
Why a Blockchain Needs an Oracle
A Solana program can read anything on Solana and nothing outside it. It has no internet access, cannot call an API and has no way of knowing what SOL trades for on a centralised exchange or what the price of gold is. Any application whose logic depends on an external price — lending protocols deciding when to liquidate, perpetuals settling positions, stablecoins checking collateral — needs that number delivered on-chain by someone.
That is the oracle’s job, and it is a genuinely hard trust problem. Whoever supplies the number can, in principle, supply a wrong one, and a wrong price triggers liquidations that should not happen. Pyth’s answer is sourcing from firms that trade the assets themselves and publishing a confidence interval alongside each price, so a consuming application can reject an update that is too uncertain rather than acting on a single unqualified figure.
Push Versus Pull
| Push oracle | Pull oracle (Pyth) | |
|---|---|---|
| Update trigger | Time or price deviation | Application requests it |
| Who pays | The oracle network | The consuming transaction |
| Cost scaling | With time and number of feeds | With actual usage |
| Freshness | As of the last push | As of the moment requested |
| Number of feeds viable | Limited by cost | Very large |
The economics explain the design choice. A push oracle pays to update every feed on a schedule regardless of whether anyone uses it, which caps how many assets it can cover. Pull moves that cost to the moment of use, so maintaining thousands of feeds becomes viable — which matters on a chain where new assets appear constantly.
The Delivery Change in 2026
Getting prices from Pythnet to other blockchains requires a cross-chain messaging layer, and that role was filled by Wormhole. After 31 July 2026 it is replaced by a five-router arrangement requiring a 3-of-5 quorum.
The reasoning is straightforward risk reduction: a single messaging layer is a single point of failure for every downstream protocol relying on those prices, and a quorum across independent routers means no one of them can deliver a bad update alone. It is the same architectural instinct as multi-client validator sets and multisig treasuries — replace one trusted component with several that must agree.
Where Oracles Do Not Reach
Memecoins generally have no oracle feed and do not need one. A memecoin’s price is whatever its liquidity pool says it is — that number is already on-chain, computed from the pool balance, so there is nothing external to import. Oracles matter for assets where an off-chain reference price exists and where a protocol needs it to make decisions.
The relevance to a memecoin trader is indirect but real. Lending protocols and perpetuals that use oracle prices are where leverage lives, and leveraged positions liquidating is a major driver of the sector-wide crashes that kill tokens in waves. An oracle failure or a delayed update during volatility cascades into forced selling that reaches everything, including assets with no oracle exposure of their own.
Reading Prices Without an Oracle
DOLAN Duck ($DOLAN) has a fixed 98.3M supply and roughly 10,700 holders, and its price is a pure pool calculation — no oracle publishes it and none needs to. The practical consequence is that any price you see for it on DEXScreener or a wallet is derived from pool balances, which means it is exactly as reliable as the pool is deep. For listed assets with oracle coverage, CoinGecko aggregates across venues and gives a more robust figure. Knowing which kind of price you are looking at is the useful takeaway: an oracle price reflects a broad market, a pool price reflects one pool, and confusing the two is how people misjudge what a market cap figure actually represents.
Pyth Network is an oracle that brings real-world prices on-chain, sourced from over 120 publishers including trading firms like Jane Street, Cumberland and Wintermute, and venues such as CBOE.
A push oracle updates feeds on a schedule whether or not anyone reads them. A pull oracle delivers an update when an application requests it, so cost scales with usage rather than with time.
Because programs cannot access anything outside the blockchain. Any protocol whose logic depends on an external price — lending, perpetuals, stablecoins — needs that number delivered on-chain by someone.
No. A memecoin’s price is computed from its liquidity pool and is already on-chain, so there is nothing external to import. Oracles matter for assets with an off-chain reference price.
Cross-chain delivery moved from Wormhole to a five-router arrangement requiring a 3-of-5 quorum after 31 July 2026, removing a single point of failure for downstream protocols.
It expresses uncertainty around the price. A consuming protocol can reject an update whose confidence is too wide, rather than acting on a single unqualified number during volatile conditions.
Indirectly. Lending and perpetuals protocols rely on oracle prices, and forced liquidations there drive sector-wide selling that reaches tokens with no oracle exposure of their own.