Pyth Network is a decentralized pull-based oracle that delivers real-time price data from institutional sources to 50+ blockchains. Unlike traditional push oracles that post prices on a schedule, Pyth lets DeFi protocols request the freshest price exactly when they need it — paying a micro-fee per update. This guide explains how Pyth works, what makes its architecture different, how PYTH token staking secures the network, and where the risks lie.
Updated April 2026 · 12 min read
Pyth Network is a decentralized oracle protocol that connects first-party financial data — prices from exchanges, trading firms, and market makers — directly to smart contracts on 50+ blockchains. Launched on Solana in April 2021, Pyth was incubated by Jump Crypto and has since grown into one of the two dominant oracle networks in DeFi alongside Chainlink.
What sets Pyth apart is its pull-based delivery model and the quality of its data sources. Rather than relying on third-party data aggregators, Pyth gets prices from the institutions that actually make markets: Jane Street, CBOE, Wintermute, Binance, OKX, Susquehanna, Two Sigma, LMAX, and dozens more. Each publisher submits its own price and confidence interval, and the protocol aggregates them into a single feed.
Traditional oracles like Chainlink use a push model: node operators post price updates on-chain at fixed intervals (e.g., every heartbeat or when price deviates by a threshold), regardless of whether any dApp consumes the update. This is reliable but expensive — every update costs gas, and protocols pay for freshness they may not need.
Pyth flips this with a pull model. Data publishers continuously stream prices to Pythnet, an application-specific Solana fork that acts as a price accumulator. Prices stay off-chain until a dApp on any target chain explicitly requests the latest update. At that point, a Wormhole-attested price is delivered on-chain and the requesting protocol pays a micro-fee (typically <$0.01).
The pull model offers three advantages. First, cost efficiency: protocols only pay for updates they actually use. Second, freshness: because updates are on-demand, a dApp can always get a price that is at most a few hundred milliseconds old, rather than waiting for the next scheduled push. Third, scalability: Pyth can support thousands of feeds across dozens of chains without the gas overhead of pushing each one.
The trade-off is that integrating protocols must trigger the price update themselves (or rely on a keeper/relayer). If no one calls for an update, the on-chain price goes stale. This makes pull oracles best suited for protocols with active users and transaction flow — like perp DEXs, lending markets, and aggregators — where every user action naturally triggers a fresh read.
Pythnet is the backbone of Pyth’s cross-chain architecture. It’s a proof-of-authority Solana fork (running the Solana validator software) dedicated exclusively to price aggregation. Publishers submit prices to Pythnet every ~400ms, and the network computes aggregated prices with confidence intervals using a stake-weighted median.
To move prices from Pythnet to target chains (Ethereum, Arbitrum, Base, BNB Chain, Sui, Aptos, etc.), Pyth relies on Wormhole — the cross-chain messaging protocol. Wormhole Guardians attest to Pythnet price updates, producing a signed Verified Action Approval (VAA). Any party on the target chain can submit this VAA to the Pyth contract to update the on-chain price.
This architecture means Pyth can serve any chain that Wormhole supports, and new chain integrations are relatively lightweight — deploy a Pyth receiver contract and the full feed catalog is immediately available. This is how Pyth expanded to 50+ chains without needing to bootstrap separate node operators for each.
Pyth’s competitive advantage over most oracles is first-party data. While Chainlink node operators typically scrape prices from public APIs (CoinGecko, CoinMarketCap, exchange APIs), Pyth’s publishers are the exchanges and trading firms themselves. They contribute prices derived from their own order books and trading activity.
The publisher roster includes CBOE, Jane Street, Susquehanna (SIG), Two Sigma, Wintermute, Virtu Financial, LMAX, Flow Traders, Binance, OKX, Bybit, KuCoin, and dozens more. This mix of TradFi and crypto-native firms gives Pyth access to data sources that simply aren’t available to third-party aggregation oracles.
Each publisher submits a price and a confidence interval (how certain they are about that price). Pythnet aggregates these into a single composite using a stake-weighted median that automatically filters outliers. If one publisher submits an extreme value, the median-based approach means it has minimal effect on the final price — unless a majority of publishers are corrupted.
The confidence interval is published alongside the price, which is a feature few other oracles offer. Protocols like Kamino and Drift can use the confidence band to widen liquidation margins during volatile periods or pause operations if confidence is too low — essentially getting a built-in volatility signal for free.
Oracle Integrity Staking is Pyth’s mechanism for aligning economic incentives with data accuracy. It introduces a staking and slashing layer on top of the publisher network.
Data publishers can stake PYTH tokens as collateral behind their price feeds. If their submissions consistently match the aggregated price (within tolerance), they earn staking rewards. If their data deviates beyond acceptable bounds, their stake can be slashed. This creates a direct financial cost for inaccurate data.
PYTH token holders who are not publishers themselves can delegate their stake to specific publishers, earning a share of the publisher’s rewards while also bearing slashing risk. This creates a reputation market: publishers with strong track records attract more delegated stake, earn more rewards, and have more skin in the game.
PYTH is the governance and utility token of Pyth Network with a max supply of 10 billion tokens. It was airdropped to early users and ecosystem participants in late 2023 and is used for governance voting, Oracle Integrity Staking, and fee parameter management.
85% of PYTH tokens were initially locked, vesting across four cliffs at 6, 18, 30, and 42 months after launch. A major unlock of 2.13B PYTH (~21.3% of max supply) is scheduled for May 2026. A DAO proposal has been put forward to delay this unlock by six months to allow a comprehensive tokenomics review — a sign the community is actively managing supply dynamics.
Pyth governance operates on 7-day epochs starting Thursdays at 00:00 UTC. To vote, you must stake PYTH tokens; staked tokens enter a warm-up period and become vote-eligible the following epoch. Governance controls feed listings, fee structures, reward parameters, and protocol upgrades. Proposals require a minimum stake to submit and run for two weeks.
Every on-chain price update costs a micro-fee paid by the protocol or user requesting it. These fees accrue to the Pyth DAO treasury. As adoption grows and more protocols pull updates, fee revenue scales without requiring more infrastructure spend — a business model that resembles an API subscription but at the protocol layer.
Pyth and Chainlink are the two dominant oracle networks but they serve different niches. Here’s how they compare across key dimensions:
| Dimension | Pyth Network | Chainlink |
|---|---|---|
| Model | Pull (on-demand) | Push (scheduled / threshold) |
| Data sources | First-party (exchanges, market makers) | Third-party (API aggregation) |
| Update latency | ~400ms (Pythnet cadence) | 1-60s (depending on feed) |
| Chains | 50+ (via Wormhole) | 30+ (native deployments) |
| Feeds | 2,800+ (crypto, equities, FX, commodities) | 1,000+ (crypto-heavy) |
| Confidence data | Yes (published per feed) | No (single price point) |
| TVS | ~$5.5B | ~$20B+ |
| Best for | Speed-sensitive DeFi (perps, lending) | Enterprise, cross-chain messaging |
In practice, many protocols use both. Chainlink’s longer track record and battle-tested push feeds make it the conservative choice for high-TVL Ethereum protocols, while Pyth dominates on Solana and is rapidly gaining ground on EVM L2s where its lower-cost pull model is a natural fit.
Oracle infrastructure is among the most critical — and most attacked — layers of DeFi. Here are the key risks to understand with Pyth:
If a majority of weighted publishers for a given feed coordinated to submit false prices, the aggregated output would be manipulated. OIS slashing raises the cost of this attack, but the risk is non-zero — especially for less-liquid feeds with fewer publishers.
Cross-chain price delivery depends entirely on Wormhole. Wormhole suffered a $325M exploit in February 2022 (since replenished by Jump). While security has been significantly hardened since, any Wormhole compromise could affect Pyth price delivery to non-Solana chains. Learn more in our cross-chain bridges guide.
During extreme market events (flash crashes, exchange outages), publishers may widen confidence intervals or go offline. If enough publishers drop, the aggregated confidence interval may become too wide for protocols to use, potentially halting liquidations or trades at the worst possible moment.
The May 2026 unlock releases 2.13B PYTH tokens (21.3% of max supply). Even with the DAO proposal to delay, the overhang creates uncertainty. Large unlocks can depress token price, which in turn reduces the economic security of OIS if publisher stakes lose value.
Pythnet is a proof-of-authority chain — its validators are permissioned. While this allows for high throughput and low latency, it means price aggregation itself is not fully decentralized. A coordinated validator failure or censorship on Pythnet would disrupt all downstream feeds.
Pyth Network is a decentralized pull-based oracle that delivers real-time price feeds for crypto, equities, forex, and commodities to 50+ blockchains. Data is sourced directly from institutional publishers like Jane Street, CBOE, Wintermute, and Binance, and aggregated on Pythnet using a stake-weighted median.
Chainlink pushes prices on-chain at fixed intervals. Pyth uses a pull model where prices are published off-chain to Pythnet and only delivered on-chain when a protocol requests an update. This makes Pyth faster and cheaper per update, while Chainlink offers more battle-tested infrastructure and a push cadence that some protocols prefer.
OIS is Pyth’s mechanism for securing price accuracy through economic incentives. Publishers and PYTH token holders stake tokens as collateral. Accurate data earns staking rewards; inaccurate data triggers slashing. This aligns the financial interests of data providers with oracle reliability.
PYTH is used for governance voting on fee structures, feed listings, and reward parameters. It’s also staked in Oracle Integrity Staking — either directly by publishers or delegated by token holders to earn yield while backing data accuracy.
As of April 2026, Pyth delivers over 2,800 price feeds spanning crypto tokens, US equities, forex pairs, commodities, and CME index futures. Coverage is expanding rapidly — 88 new feeds were added in a single week in March 2026.
Key risks include publisher collusion on low-liquidity feeds, Wormhole bridge dependency for cross-chain delivery, confidence interval failures during extreme volatility, the May 2026 token unlock overhang (2.13B tokens), and Pythnet validator centralization.