Pyth Network MCP Server 1.1 Adds Tick-Level Price History and New Market Data Feeds
Pyth Network has released version 1.1 of its Pyth Pro Model Context Protocol (MCP) Server, expanding the market data tools available to AI agents. Announced on October 15, 2026, the update introduces tick-level price history, feed availability checks, simpler asset-pair queries and additional financial data. The changes aim to make it easier for AI applications to retrieve and interpret market information across multiple asset classes.
Tick-Level History and Simpler Market Data Queries
One of the main additions is the get_price_range tool, which returns individual price updates within a selected window of up to 60 seconds. This gives AI agents access to more granular information than simply comparing the starting and ending prices of a period. For example, an agent investigating market activity around a company earnings announcement or an interest-rate decision can examine the sequence of price updates during that window.
The feature could be useful for developers building market research assistants, trading analytics tools and automated monitoring systems. Short-lived price movements can disappear from view when data is reduced to a few snapshots, making granular history valuable when examining how markets respond to specific events. However, the availability of detailed price data does not itself guarantee accurate predictions or profitable trading decisions.
Pyth Pro MCP Server 1.1 also introduces the get_symbols tool for users with a Pro API key. It lists available feeds and identifies which ones are covered by the user’s subscription plan. According to Pyth, this allows an AI agent to distinguish between a feed that falls outside the user’s plan and one that is not live before requesting its price.
Another change simplifies how agents request market data. Instead of looking up a complete Pyth feed identifier, users can provide familiar pairs such as BTC/USD. This functionality extends across cryptocurrencies, foreign exchange, equities and metals. If a requested pair matches multiple feeds, the server indicates the ambiguity rather than silently treating the request as a unique match.
The release also expands the range of information that agents can retrieve. New or exposed data includes funding rates, exponential moving average (EMA) prices, market sessions and update channels with intervals as short as 50 milliseconds. Interest-rate and net asset value (NAV) feeds are also newly available through the server, extending its coverage beyond the asset classes commonly associated with crypto-focused data tools.
Better Error Handling and Broader AI Integration
Pyth has also updated the server’s error reporting. When a requested feed or data channel cannot be accessed, the agent can receive a more specific explanation, such as an invalid token, a feed outside the subscription plan, an inactive feed or an unsupported channel. More precise responses can help developers identify configuration problems and reduce unnecessary retries.
Setup is designed to remain straightforward. Users add their Pyth Pro token to their MCP client configuration, after which the client can use the credentials automatically for subsequent calls. MCP provides a standard way for AI applications to interact with external tools and data sources, allowing developers to connect supported market data services to compatible agent workflows.
The broader significance of the update lies in how AI systems access financial information. An agent that can request historical price updates, identify available feeds and retrieve several types of market data has more tools for answering research questions without requiring a developer to manually translate every request into a technical feed identifier. That convenience could support more flexible market analysis applications.
Still, the practical value of these tools depends on data quality, feed coverage, latency, subscription access and how developers design their applications. A 50-millisecond update channel should not be confused with a guarantee that every part of an AI trading system operates at that speed. Nor does the release establish that the server independently executes trades or generates investment recommendations.
For the Pyth ecosystem, version 1.1 represents an expansion of the interface between financial data and AI agents. Its combination of granular price history, wider asset coverage and clearer access errors gives developers more ways to build market-aware applications. Adoption will depend on whether those capabilities solve real problems for users building research, monitoring and trading-related software.
About The Author
