Crypto Gloss

Pyth Network

Pyth Network solves a core DeFi infrastructure problem: price oracles. Traditional oracles like Chainlink use a push model — data is posted to every chain at regular intervals, charging fees regardless of whether anyone queries the price. Pyth uses a pull model — first-party data from 95+ professional trading firms and market makers is aggregated off-chain, and users fetch and post it only when needed. This dramatically reduces costs and enables latency as low as 400ms — far faster than block-time-based oracles. Originally built for Solana, Pyth now serves as the oracle infrastructure for hundreds of DeFi protocols across 50+ EVM and non-EVM chains.

The Oracle Problem

For DeFi to work, smart contracts need reliable real-world prices (e.g., “what is ETH worth right now?”). Bad oracle prices enable:

  • Manipulation attacks: Flash loans combined with oracle price manipulation to drain lending protocols
  • Stale prices: Prices that don’t reflect volatile market moments, leading to insufficient liquidations
  • Front-running: Bots exploiting the gap between market price and oracle-reported price

Classic oracle approaches:

  • Centralized: Trust a single source (Coinbase price API) — single point of failure
  • Push decentralized (Chainlink): Multiple validators fetch + combine prices, post on schedule
  • AMM-derived (Uniswap TWAP): Use DEX pool ratios as prices — manipulable with large trades

Pyth’s innovation: first-party, pull-based, sub-second pricing from institutions that have direct market access.

How Pyth Works

The following sections cover this in detail.

Publishers

Pyth’s data comes from 95+ first-party publishers — the entities that actually see and make markets:

  • Trading firms: Jane Street, Jump Crypto, Cumberland, Flow Traders, Auros
  • Market makers: Wintermute, B2C2, Galaxy Digital
  • Exchanges: Binance, Coinbase, OKX, Bybit, Kraken
  • Crypto-native MMs: Alameda Research (defunct), GSR, Amber Group

Why first-party matters: A trading firm posting sETH/USDC prices is posting their own actual mid-market price — the most accurate number they have, because they trade against it. Compare to a third-party oracle aggregator re-processing exchange APIs with potential delays.

Aggregation

Each publisher submits their price + confidence interval (e.g., “BTC is $68,000 ± $50”). Pyth’s aggregation algorithm:

  • Takes the median price across publishers (outlier resistant)
  • Computes a confidence interval from the distribution

The confidence interval is key: protocols can decide to refuse execution if the confidence interval is too wide (indicates market turbulence or low liquidity).

Pull Oracle Model

“`

Publisher → Wormhole Guardian Network → Pythnet (Solana-based accumulator)

User requests price → Fetch signed price from Pythnet → Post attestation to target chain → Protocol reads verified price

“`

Crucially, the user (or protocol) pays the L1/L2 gas to post the price. No continuous broadcast cost for Pyth. Fast enough to be queried per-transaction.

Comparison:

  • Chainlink push: Price updated every few minutes or when deviation > 0.5% — always on-chain but stale
  • Pyth pull: Price updated 400ms ago, fetched fresh each transaction — fresher but requires user gas

Cross-Chain via Wormhole

Pyth uses Wormhole’s Guardian Network (19 validators) to sign and forward price messages from Pythnet to all supported chains. This means:

  1. Security reliance on Wormhole’s security model
  2. Messages arrive on any chain in ~2-3 seconds (Wormhole messaging time)
  3. Price attestation is cryptographically signed — can’t be forged

On the target chain, the Pyth receiver contract verifies the Wormhole guardian signatures before accepting the price.

PYTH Token

Launch: November 2023 — one of the largest airdrops of the year

Airdrop recipients: Users who had traded on protocols using Pyth price feeds

Total supply: 10 billion PYTH

Utility:

  • Governance: Vote on protocol parameters (minimum publisher count, confidence intervals, oracle fees)
  • Staking for governance: Lock PYTH to participate in votes weighted by stake
  • Future: Potential publisher rewards funded by staking

Airdrop mechanics:

  • 6% of supply (600M PYTH) to DeFi users across 27 protocols
  • Allocated per protocol based on usage; users could claim if they’d interacted with any Pyth-powered protocol
  • Significant value — PYTH traded around $0.40 at TGE (all-time value: ~$0.90)

Pyth Price Feeds

Number of feeds: 450+ across:

  • Crypto: All major tokens + long-tail assets
  • Equities: US stocks (AAPL, TSLA, GOOGL, SPY)
  • FX: USD/EUR, USD/JPY, etc.
  • Commodities: Gold, silver, oil
  • Rates: US Treasury yields

The non-crypto feeds are particularly important for emerging RWA (Real World Asset) protocols that need equity prices on-chain.

Confidence Intervals in Practice

A Pyth feed for an illiquid token might report:

“`

Price: $1.00 ± $0.15

“`

This huge ±15% confidence means publishers disagree significantly. Protocols like Drift (Solana perps) use confidence intervals to pause liquidations during high volatility — preventing bad debt from stale/uncertain prices.

Ecosystem Usage

Solana (dominant):

  • Jupiter: Uses Pyth for swap price checking
  • Drift Protocol: Perps DEX; entire price engine based on Pyth ORACLE pricing
  • Mango Markets: Former flagship Solana DEX (hacked); used Pyth (the hack exploited confidence interval weakness)
  • Kamino Finance: Solana lending, Pyth prices
  • Jito: MEV-related price checking

EVM:

  • Synthetix: Pyth as primary oracle since v3
  • GMX v2: Multi-oracle using Chainlink + Pyth
  • Kwenta: Synthetix perps frontend
  • Dozens on Base, Arbitrum, Optimism

Pyth vs. Chainlink

Pyth Chainlink
Model Pull (on-demand) Push (scheduled)
Latency ~400ms Minutes to hours
Data sources First-party institutions Third-party node operators
Coverage 450+ feeds incl. equities/FX 1000+ crypto feeds
Cost User pays per query Protocol pays subscription
Manipulation risk Harder (large institutions, median) Possible via node coordination
EVM dominance Growing Dominant

Neither is universally superior — Chainlink’s push model is better for passive/long-tail needs; Pyth’s pull is better for high-frequency on-chain execution (perps, liquidations).

How to Use Pyth Data

For developers:

“`javascript

// Fetch and verify price update from Pyth

const connection = new PriceServiceConnection(“https://hermes.pyth.network”);

const priceFeedId = “0xe62df6c8b4a85fe1a67db44dc12de5db330f7ac66b72dc658afedf0f4a415b43”; // BTC/USD

const priceFeeds = await connection.getLatestPriceFeeds([priceFeedId]);

const priceUpdateData = priceFeeds[0].getVAA();

// Post priceUpdateData to chain → read verified price

“`

For end users:

Pyth is infrastructure — users interact with protocols using Pyth, not Pyth directly. If you’re trading on Drift or borrowing on Morpho, Pyth is working in the background.

Get tokens for DeFi protocols via . Secure assets with .

Social Media Sentiment

Pyth is well-regarded on CT as the high-performance oracle for Solana and fast chains. The PYTH token has mixed sentiment — strong protocol utility but price performance has trailed broader oracle sector expectations. Tech-focused CT accounts celebrate Pyth’s sub-second latency and institutional publisher roster. Cross-chain expansion is a positive narrative for the protocol.

Last updated: 2026-04

Related Terms

Sources

Al-Bassam, M., Sonnino, A., Buterin, V., & Khabbazian, M. (2019). Fraud and Data Availability Proofs: Maximising Light Client Security and Scaling Blockchains with Dishonest Majorities. arXiv.

Peterson, J. (2015). Augur: A Decentralized, Open-Source Platform for Prediction Markets. arXiv.

Liu, B., Szalachowski, P., & Zhou, J. (2020). A First Look into DeFi Oracles. IEEE.

Eskandari, S., Salehi, M., Gu, W. C., & Clark, J. (2021). SoK: Transparent Dishonesty — Front-Running Attacks on Blockchain. Financial Cryptography.

Chainlink Whitepaper. (2017). A Decentralized Oracle Network. SmartContract.com.