Pyth inverts the oracle model. Traditional oracles like Chainlink use a push model, where data is broadcast to all contracts, creating redundant on-chain gas costs. Pyth’s pull model requires data consumers to explicitly request and pay for the specific price updates they need, eliminating waste.
the-state-of-web3-education-and-onboarding
Blog
Why Pyth's Pull Model Changes the Oracle Economics Game
Pyth Network's pull-based oracle inverts the gas cost model, shifting update control to dApps. This enables high-frequency trading and micro-transactions but introduces critical liveness dependencies. We break down the economic trade-offs and the new attack vectors it creates.
introduction
THE PULL
Introduction
Pyth's pull oracle model inverts the economic incentives and technical architecture of data delivery, creating a more efficient and secure market.
This creates a direct market. The model aligns incentives between data publishers and consumers, mirroring the efficiency of off-chain systems like The Graph for queries or UniswapX for intents. Consumers pay for value received, and publishers earn fees for data actually used.
The economic shift is profound. Push models create a tragedy of the commons where a few users subsidize data for many. Pyth’s pull model makes data a private good, forcing applications like perpetual DEXs and lending protocols to internalize their oracle costs, leading to more sustainable protocol economics.
Evidence: Since launch, Pyth has secured over $100B in on-chain value, demonstrating that major DeFi protocols like Synthetix and Venus trust its pull-based data for critical financial operations.
key-trends
ORACLE COST DYNAMICS
The Push vs. Pull Economic Divide
Traditional push oracles like Chainlink charge protocols for constant data delivery, creating misaligned incentives and wasted gas. Pyth's pull model flips the script.
01
The Problem: Sunk Cost of Push Updates
Legacy oracles push data on every block, forcing protocols to pay for unused updates and subsidize competitors' data consumption. This creates a tragedy of the commons in cost allocation.
- Wasted Gas: Protocols pay for ~70-80% of updates they never use.
- Misaligned Incentives: Data consumers have no direct control over update frequency or cost.
~80%
Wasted Updates
Sunk Cost
Economic Model
02
The Solution: Pyth's On-Demand Pull
Pyth Network stores price feeds on-chain; consumers pull data only when needed, paying the gas themselves. This aligns cost with actual usage, like a utility bill.
- Pay-Per-Use: Fees are ~50-90% lower for low-frequency applications.
- Consumer Sovereignty: Each protocol optimizes for its own latency and cost requirements.
-90%
Potential Cost Save
On-Demand
Pricing
03
The Consequence: Unbundling Oracle Stacks
The pull model separates data publishing from delivery, enabling new infrastructure like Flashbots SUAVE for MEV-aware updates and specialized puller networks like Chronicle's Scribe. This mirrors the UniswapX and Across model for intents.
- Specialization: Publishers focus on data quality, pullers on delivery efficiency.
- MEV Integration: Updates can be bundled with user transactions, hiding latency.
Modular
Stack Design
MEV-Aware
New Frontier
04
The Trade-off: Latency vs. Cost
Pull models introduce a ~100-400ms latency penalty for the on-chain pull operation. This creates a new design space where protocols choose their own cost-latency frontier, unlike the one-size-fits-all push approach.
- High-Freq Trading: May still prefer push for sub-second updates.
- Lending/Options: Ideal for pull, prioritizing cost savings over millisecond latency.
~400ms
Pull Latency
Design Choice
Protocol Control
05
The Competitor: Chainlink's CCIP & Hybrid Future
Chainlink is responding with CCIP, a cross-chain messaging layer that could enable hybrid models. The endgame isn't push or pull, but oracles as verifiable data layers where both models coexist, similar to EigenLayer's restaking for security.
- Data Layer: Oracle networks become settlement layers for verifiable information.
- Hybrid Models: Push for hotspots, pull for long-tail assets.
Hybrid
Future State
Data Layer
Endgame
06
The Verdict: Economic Alignment Wins
Pyth's model proves that economic alignment drives adoption. By making the consumer the payer, it eliminates subsidy wars and forces infrastructure to compete on efficiency. This is the same principle behind Uniswap v4 hooks and EigenLayer restaking: granular control over cost and performance.
- Adoption Proof: $10B+ in on-chain value secured.
- Paradigm Shift: From 'data as a subscription' to 'data as a commodity'.
$10B+
Secured Value
Aligned
Incentives
deep-dive
THE ECONOMIC SHIFT
The High-Frequency Frontier: Use Cases Unleashed
Pyth's pull model inverts oracle economics, enabling new high-frequency applications by shifting cost and latency burdens to the consumer.
The cost burden shifts from the publisher to the consumer. In traditional push oracles like Chainlink, publishers pay to push data on-chain, creating a latency-cost trade-off. Pyth's pull model makes consumers pay to pull data, freeing publishers to broadcast updates at sub-second speeds without incurring gas costs.
This enables high-frequency applications previously impossible. Perpetual DEXs like Hyperliquid and Drift Protocol require price updates for every trade and liquidation. The pull model's low-latency data stream allows these protocols to operate with CEX-like efficiency, supporting millisecond-level market operations.
It creates a direct performance incentive. Publishers compete on data quality and speed, not just on who can afford the most gas. This market dynamic, visible on networks like Solana and Sui, continuously drives down update latency as publishers like Jane Street and Jump Crypto optimize their feeds.
Evidence: Pyth data updates occur every 300-400 milliseconds on Solana, compared to Chainlink's typical 1-2 second heartbeat on Ethereum L2s. This order-of-magnitude difference is the foundation for the sub-second liquidations and tight spreads seen in leading on-chain perpetuals markets.
PULL VS. PUSH ARCHITECTURE
Oracle Model Comparison: Economic & Risk Profile
A first-principles breakdown of how oracle data delivery models fundamentally alter capital efficiency, risk vectors, and protocol economics.
| Feature / Metric | Pull Model (Pyth) | Traditional Push Model (Chainlink) | Hybrid/On-Demand |
|---|---|---|---|
Primary Data Delivery | On-Demand Consumer Pull | Continuous Provider Push | Variable |
Capital Efficiency (Staker/Node) | High (Capital not locked per feed) | Low (Capital locked per feed) | Medium |
Latency (Update to On-Chain) | Sub-second (Consumer-triggered) | Block-time bound (e.g., 12-30 sec) | Variable |
Cost Model | Pay-per-update (User pays gas) | Subsidized by protocol/DAO | Mixed |
Data Freshness Guarantee | Immediate (User-driven) | Probabilistic (Heartbeat-based) | Conditional |
Sybil Resistance Mechanism | Stake-slashing on Pythnet | Stake-slashing on Mainnet | Depends on implementation |
Primary Risk Vector | User transaction failure | Oracle downtime/outages | Complexity |
Example Protocols | Pyth Network | Chainlink Data Feeds | API3, DIA |
risk-analysis
ORACLE ARCHITECTURE SHIFT
The Liveness Risk Vector: What Breaks in a Pull World
Pyth's pull-based model inverts the traditional oracle risk model, transferring liveness responsibility from the publisher to the consumer.
01
The Problem: The Push Oracle Bottleneck
Traditional oracles like Chainlink push data on-chain at a fixed cadence, creating a single point of failure. The protocol bears the cost and risk of every update, even when unused.
- Liveness Risk: A publisher outage halts all downstream protocols.
- Economic Waste: Paying for ~1M+ updates/day to serve a fraction of that demand.
- Latency Ceiling: Updates are batch-bound, creating ~1-10 second delays.
~1-10s
Update Latency
1M+/day
Redundant Updates
02
The Solution: On-Demand Data Pulls
Pyth's model makes data consumers (e.g., Perpetual DEXs, lending markets) responsible for pulling the latest price when they need it. The oracle network commits a price on a low-latency Pythnet, and consumers pull a signed price attestation on-demand.
- Risk Transfer: Liveness risk shifts from publisher to the application pulling the data.
- Cost Efficiency: Pay only for the data you consume, slashing gas fees.
- Sub-Second Finality: Pulls can be integrated into a user transaction for ~400ms latency.
~400ms
Pull Latency
-90%
Gas Cost (vs. push)
03
The Consequence: New Economic & Security Primitives
This flips oracle economics from a cost center to a performance lever. Protocols now compete on update freshness and cost efficiency.
- MEV Integration: Pulls enable JIT liquidity and oracle-frontrunning protection, similar to UniswapX.
- Protocol-Level SLAs: Applications can define their own liveness guarantees and redundancy (e.g., fallback to Chainlink).
- Publisher Specialization: Data providers compete on latency and attestation security, not just uptime.
JIT
Liquidity Enabled
SLA-Driven
Protocol Design
04
The Trade-off: Who Bears the Tail Risk?
The pull model's elegance has a dark side: it externalizes extreme congestion risk. In a network-wide black swan event, every protocol scrambles to pull data simultaneously.
- Gas Auction Risk: Critical updates could trigger gas price wars during volatility.
- No Default Safety Net: If no one pulls, there is no price. This demands new keeper network designs.
- Asymmetric Burden: The largest protocols (e.g., Aave, Compound) become de facto liveness backstops for the ecosystem.
Gas Wars
Tail Risk
Keeper Nets
Required
future-outlook
THE ECONOMICS
Hybrid Futures and the End of One-Size-Fits-All
Pyth's pull model inverts oracle economics, enabling specialized data feeds that render monolithic competitors obsolete.
Pull model inverts fee capture. Traditional oracles like Chainlink charge protocols for data pushes, creating a rent-seeking model. Pyth's pull model lets applications pay only for the data they consume, aligning costs directly with utility.
Specialization destroys generalization. This enables hyper-specialized data feeds for niche assets or low-latency derivatives, a market that push oracles cannot serve profitably. It's the Uniswap V3 moment for oracles, where concentrated liquidity beats generic pools.
Protocols become data publishers. Projects like Synthetix Perps or Drift Protocol can now monetize their own proprietary price streams directly, turning a cost center into a revenue stream and improving data freshness.
Evidence: Pyth's Solana feed updates every 400ms for ~$0.0001 per pull, a cost structure that makes high-frequency on-chain trading viable, a feat impossible with push-model gas economics.
takeaways
ORACLE ECONOMICS
TL;DR for Protocol Architects
Pyth's pull-based data delivery inverts the traditional oracle cost model, shifting latency and gas risk from the protocol to the user.
01
The Problem: Push-Model Inefficiency
Traditional oracles like Chainlink push updates on-chain, forcing protocols to pay for unused data and bear the latency/gas risk of every update. This creates a fixed, recurring cost regardless of actual usage, with ~15-30 second update latencies.
- Protocols subsidize all users for data, even inactive ones.
- Gas volatility risk is held on the protocol's balance sheet.
- Update frequency is a trade-off between cost and freshness.
~30s
Typical Latency
Fixed Cost
Pricing Model
02
The Solution: On-Demand Pull Oracle
Pyth stores price attestations on a low-cost publisher network. Users (or their apps) pull the latest signed price on-demand via a permissionless on-chain program, paying only for the data they consume.
- Users pay for their own data, aligning cost with usage.
- Sub-second finality is possible as the latest price is always available.
- Protocols eliminate recurring oracle gas bills and latency management.
<1s
Price Finality
Pay-per-Call
Cost Model
03
The Impact: Redefined Protocol Economics
This shifts oracle costs from a protocol-level CAPEX to a user-level OPEX, fundamentally changing treasury management and product design for DeFi protocols like perpetuals DEXs and lending markets.
- Treasury runway extends as fixed data costs vanish.
- New product designs become viable (e.g., ultra-low-fee perpetuals).
- Competitive moat shifts to user experience and execution, not who can afford the oracle bill.
$0
Protocol Fixed Cost
UX Advantage
Competitive Edge
04
The Trade-off: Relayer Incentives & Composability
The pull model requires a decentralized network of relayers to post prices on-chain. This introduces a new incentive layer that must be robust against MEV and ensure data availability for composable calls from protocols like Uniswap or Aave.
- Relayer profitability must be sustained via fees and MEV opportunities.
- Data freshness guarantees are critical for flash loan attacks.
- Composability latency is now a function of relayer performance, not a fixed schedule.
New Layer
Relayer Network
MEV-Driven
Incentive Design
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall