What Is Pyth Network? The Pull Oracle Feeding 100+ Chains, Explained

Every DeFi protocol needs to know one thing before it can do anything else: what is this asset worth right now? That single question is the oracle problem, and Pyth Network answers it differently from almost everyone else. Instead of paying node operators to scrape public price APIs and push the result on-chain, Pyth convinced the firms that actually trade the assets — Jane Street-tier market makers, exchanges, proprietary trading desks — to publish their own prices directly. Then it flipped the delivery model: rather than constantly writing prices to every chain, it keeps them off-chain and lets apps pull the freshest number only when they need it. The result is the fastest, most widely deployed first-party oracle in crypto, and Chainlink's most credible challenger. Here's how it works, what the PYTH token actually does, and where the real risks sit.

What Pyth Network is and who built it

Pyth Network is a decentralized oracle for real-time financial market data. It covers crypto, US and international equities, foreign exchange, commodities, ETFs and a growing list of real-world assets — over 2,000 price feeds in total — and delivers them to more than 100 blockchains. Its defining trait is first-party data: more than 120 institutions, including some of the largest market makers and exchanges in the world, publish their proprietary prices straight to the network. Most oracles are middlemen aggregating data they don't own; Pyth is a consortium of the people who make the markets.

The project's origin explains that access. Pyth was incubated inside Jump Crypto, the digital-assets arm of Chicago trading powerhouse Jump Trading, and went live on Solana in August 2021, initially with feeds for around 30 crypto assets. Jump's relationships with other trading firms were the seed capital that no pure software team could replicate. In July 2023 the core contributors spun out into an independent company, Douro Labs, led by CEO Michael Cahill, which builds much of the protocol today. Governance itself sits with the token-holding Pyth DAO, whose permissionless mainnet launched alongside the PYTH token in late 2023.

How the pull oracle actually works

This is where Pyth diverges most sharply from the Chainlink model most people picture. Follow the data through four stages:

1. Publishers submit. Each of the 120+ data providers pushes its own price for an asset, together with a confidence interval — the firm's own estimate of how uncertain that price is — to Pythnet, a dedicated application-specific blockchain built on Solana technology.
2. Pythnet aggregates. Roughly every 400 milliseconds, Pythnet combines all the individual submissions into a single aggregate price and a single aggregate confidence interval per feed. Outliers and low-confidence inputs are weighted down, so no single publisher can yank the price.
3. Wormhole attests. The signed aggregate is made available cross-chain through Wormhole, the interoperability layer. A price update becomes a signed message that any Pyth contract, on any supported chain, can verify.
4. Applications pull. Instead of Pyth writing the price to your chain on a schedule, your app fetches the latest signed update off-chain (via a service called Hermes) and posts it on-chain in the same transaction that uses it. You pay the gas only at the moment you consume a price.

That last step is the "pull" in pull oracle, and it's the whole economic argument. A push oracle must pay to write every feed to every chain on a fixed cadence, whether anyone reads it or not — which is why push oracles ration themselves to a few hundred feeds on a handful of chains. Pyth offloads that cost to the data consumer, so it can afford to offer thousands of feeds on dozens of chains. Adding a new chain is close to free: deploy the Pyth contract, point it at the same Wormhole signatures, and the entire feed catalogue is instantly available.

The confidence interval is the underrated feature

Most oracles hand you a single number and leave you to trust it. Pyth hands you a number and a spread. If publishers broadly agree, the confidence interval is tight; if they're reporting wildly different prices — a sign of a thin, illiquid or actively manipulated market — the interval blows out. A protocol that integrates Pyth properly can treat that spread as a circuit breaker: if the interval widens past a threshold, pause new borrows, widen collateral haircuts, or fall back to a secondary feed. It's a native risk-management signal that push oracles simply don't provide, and it's one of the strongest arguments for using Pyth in leveraged products.

The PYTH token and Oracle Integrity Staking

Pyth has a token, and unlike some oracle tokens it does real work. PYTH launched on 20 November 2023 via a retrospective airdrop to roughly 90,000 wallets across 27 chains, with a maximum supply of 10 billion. Around 7.9 billion was circulating by mid-2026, and the supply schedule matters: the bulk of tokens unlock at 6, 18, 30 and 42 months after launch, with a substantial cliff — roughly 2.1 billion tokens — landing in May 2026 and pushing circulating supply up from about 5.75 billion. That overhang is a live consideration for anyone weighing the token rather than the technology.

PYTH has two jobs. The first is governance: staked PYTH votes on the DAO's decisions — fee parameters, new feeds, reward structures, treasury use. The second is more interesting. In 2024 Pyth launched Oracle Integrity Staking (OIS), which puts economic skin in the game behind data quality. Holders stake PYTH to specific publishers. If a publisher reports accurate prices, stakers earn rewards; if a publisher reports bad data, a slice of the stake backing it can be slashed. This turns data integrity into a bonded commitment rather than a promise — the firms supplying prices, and the people vouching for them, lose money when the feed is wrong. By the end of 2025, roughly 820 million PYTH was staked in OIS.

Pyth has also started generating actual revenue. Pyth Pro, a subscription tier aimed at institutions that want the data off-chain too, crossed $1 million in annualized recurring revenue, with a 60% revenue share flowing to the DAO treasury and funding regular PYTH buybacks (recently a couple of million tokens per month). It's early, but it's a real business model layered on top of a public good — the "Phase 3" ambition of monetizing the roughly $50 billion global market-data industry, not just DeFi.

The numbers and market position

Here's where honesty beats hype. Pyth is routinely called the number-two oracle, and by most measures that's fair — but not by every measure, and the exception is worth knowing.

• By total value secured (TVS), the DefiLlama metric of assets protected by an oracle, Pyth actually sits behind both Chainlink (tens of billions) and Chronicle (which secures several billion concentrated largely in the Sky/MakerDAO ecosystem). Pyth's TVS has ranged in the several-billion-dollar area with a single-digit market share. Chronicle's number is high because it protects a few very large protocols; it is not a broader network.
• By reach, Pyth leads the challengers decisively: 100+ chains, 2,000+ feeds, and 400+ integrations — far more than any oracle except Chainlink, whose chain coverage is broadly comparable (both are in the ~100-chain range, though exact counts vary by source).
• By data freshness and first-party sourcing, Pyth is arguably ahead of everyone. Its ~400ms aggregation and its Lazer low-latency product (millisecond-grade feeds for high-frequency trading and perps) target a niche push oracles can't serve.

So the accurate framing is: Pyth is the clear number-two oracle by network breadth, brand and first-party data, but only a single-digit-share player by raw TVS, trailing Chainlink, Chronicle and roughly on par with the other challengers. Which metric matters depends on what you're building. For a high-throughput perps DEX, Pyth's latency and chain coverage are the whole game; for a conservative lending market, Chainlink's deeper TVS track record still carries weight.

Ecosystem, integrations and the institutional turn

Pyth's home turf is derivatives. Perp and options venues like Drift and Synthetix, lending markets like Euler, and hundreds of other protocols across Solana, Ethereum, the major L2s, Aptos, Sui, TON, Sei, Berachain, Monad and more rely on Pyth for mark prices, liquidations and collateral valuation. Because a single signed update works on any supported chain, a multi-chain protocol gets consistent pricing everywhere without wiring up a different oracle per deployment.

The bigger 2025-2026 story is Pyth moving up-market into TradFi data. Two milestones stand out. First, in August 2025 the US Department of Commerce selected Pyth to verify and distribute official economic data on-chain, beginning with GDP — publishing an authoritative government statistic to blockchains, with more indicators planned. The PYTH token spiked sharply on the news. Second, in June 2026 Nasdaq moved to distribute its institutional-grade TotalView depth-of-book equity data through Pyth, putting full order-book data for Nasdaq-listed stocks on-chain. These aren't crypto-native integrations; they're signals that Pyth is positioning as neutral price infrastructure for finance broadly.

How to actually use Pyth

For a developer, integrating Pyth follows a consistent shape regardless of chain:

1. Find the feed ID for the asset you need in Pyth's price-feed registry (each feed has a stable ID across all chains).
2. Fetch the latest update off-chain from Hermes, Pyth's price service, which returns a signed update blob for your chosen feeds.
3. Post and consume it on-chain in one transaction: call updatePriceFeeds on the Pyth contract (paying a small update fee) and then read the price with getPriceNoOlderThan, which enforces a maximum staleness so you never act on an outdated number.
4. Use the confidence interval. Read it alongside the price and set a threshold; reject or defensively handle any update where the spread is too wide.

The two things a correct integration must do are enforce staleness checks (never trust a price older than N seconds) and respect the confidence interval. Nearly every oracle-related exploit in DeFi traces back to a protocol that skipped one of these — reading a stale price, or trusting a manipulated thin-market price with no sanity check. Pyth gives you the tools; using them is on the integrator.

Risks and what to watch

A clear-eyed guide names the failure modes:

• Wormhole dependency. Cross-chain delivery runs through Wormhole, which means Pyth inherits Wormhole's trust assumptions. Wormhole was exploited for roughly $325 million in February 2022 (funds were later backfilled). The bridge has been hardened since, but it remains a shared point of failure for Pyth's multi-chain distribution.
• Liveness. On 22 May 2026, Pythnet and the Hermes service suffered a multi-hour outage that halted price updates across chains — a reminder that "pull" doesn't help if there's no fresh update to pull. Protocols relying on Pyth as a sole feed felt it.
• Integration risk. The pull model pushes responsibility onto the consumer. Skip a staleness check or ignore the confidence interval and you've built an exploit, not a safeguard. This is a design tradeoff, not a Pyth bug — but it's real.
• Publisher concentration. First-party data is a strength, but the publisher set overlaps with a handful of large trading firms. OIS slashing is designed to bond that trust; whether it's been battle-tested through a genuine adversarial event at scale is still open.
• Token supply overhang. The May 2026 unlock cliff and the broader vesting schedule put persistent sell-pressure on PYTH that's independent of the protocol's technical merits.

Bottom line

Pyth Network took two contrarian bets — that oracles should carry first-party data straight from the firms that trade the assets, and that prices should be pulled on demand rather than pushed on a schedule — and both aged well. The payoff is an oracle that runs thousands of low-latency feeds across 100+ chains cheaply, ships a confidence interval most competitors don't, and has parlayed its Jump Trading pedigree into landmark institutional deals — a US Commerce Department GDP feed (awarded alongside Chainlink) and Nasdaq TotalView distribution. It is not the largest oracle by value secured — Chainlink leads and Chronicle's concentrated TVS edges it out on that one metric — but by reach, data quality and institutional momentum it is the clear challenger for the throne. Watch the Wormhole dependency, respect the confidence interval when you build on it, and treat the token's unlock schedule as separate from the network's genuine strength.

Related guides

• What Is Chainlink?
• What Is Solana?
• What Is The Graph?
• What Is Synthetix?
• Injective Guide

For related reading: what is DeFi, best Ethereum L2s, and account abstraction explained.

Not financial advice. PYTH is volatile and oracle integrations carry technical risk — always verify details on Pyth's official documentation and channels.