AI can summarize a financial news story, but it may not be able to answer this: What exactly did the price do in the 30 seconds after the news broke? What Pyth has added for agents this time is a tool for querying data.
Pyth released Pro MCP Server 1.1 on October 15. The new version can retrieve every available price update within a window of up to 60 seconds, letting researchers compare more than just the starting and ending points. For example, prices at two timestamps may be similar even though there was significant movement in between; snapshots from before and after would miss that. This is for research purposes—the data interface should not be mistaken for an automated trading or profit-making system.
The technical documentation also distinguishes between “discoverable” and “usable”: get_symbols can be used to discover data sources; if you provide your own Pro Key, it will also indicate whether your plan grants access. Only data sources with stable status can be queried for prices. Sources listed as beta or coming_soon are not necessarily readable now. For developers, this can help avoid conflating insufficient permissions, unavailable sources, and missing data into a single error.
Its practical use is to connect news retrieval with market-data review. In their official announcement on October 7, Pyth and Exa still described their work together as early testing: search finds events and materials, while structured market data adds market observations. The earlier partnership should not be presented as proof that all use cases are commercially deployed today.
My view is that the product progress is worth studying, but its value will depend on data completeness, access coverage, and efficiency in real workflows. Other market-data APIs can also support event research, so unified access may not create an exclusive advantage; having more granular data also cannot by itself prove that news caused a price move.
Collected at 04:05 Beijing time on October 16: PYTH/USDT on Binance was up about 14.82% over the rolling 24 hours, with trading volume of about 23.29 million USDT. The full 03:00–04:00 hour was up about 0.92%, with trading volume of about 835,000 USDT and a volume ratio of about 0.40. Strong intraday activity and low volume in the most recent hour can coexist. Market data does not prove the independent impact of MCP adoption or the version announcement.
Going forward, I’ll watch to see whether the features continue to be delivered, whether users can access the data they need, and whether there is evidence of actual use. If the documented capabilities do not match verifiable results, I’ll revise my assessment of the product. This review only verified official materials; I did not connect to or test the service. $PYTH
Pyth released Pro MCP Server 1.1 on October 15. The new version can retrieve every available price update within a window of up to 60 seconds, letting researchers compare more than just the starting and ending points. For example, prices at two timestamps may be similar even though there was significant movement in between; snapshots from before and after would miss that. This is for research purposes—the data interface should not be mistaken for an automated trading or profit-making system.
The technical documentation also distinguishes between “discoverable” and “usable”: get_symbols can be used to discover data sources; if you provide your own Pro Key, it will also indicate whether your plan grants access. Only data sources with stable status can be queried for prices. Sources listed as beta or coming_soon are not necessarily readable now. For developers, this can help avoid conflating insufficient permissions, unavailable sources, and missing data into a single error.
Its practical use is to connect news retrieval with market-data review. In their official announcement on October 7, Pyth and Exa still described their work together as early testing: search finds events and materials, while structured market data adds market observations. The earlier partnership should not be presented as proof that all use cases are commercially deployed today.
My view is that the product progress is worth studying, but its value will depend on data completeness, access coverage, and efficiency in real workflows. Other market-data APIs can also support event research, so unified access may not create an exclusive advantage; having more granular data also cannot by itself prove that news caused a price move.
Collected at 04:05 Beijing time on October 16: PYTH/USDT on Binance was up about 14.82% over the rolling 24 hours, with trading volume of about 23.29 million USDT. The full 03:00–04:00 hour was up about 0.92%, with trading volume of about 835,000 USDT and a volume ratio of about 0.40. Strong intraday activity and low volume in the most recent hour can coexist. Market data does not prove the independent impact of MCP adoption or the version announcement.
Going forward, I’ll watch to see whether the features continue to be delivered, whether users can access the data they need, and whether there is evidence of actual use. If the documented capabilities do not match verifiable results, I’ll revise my assessment of the product. This review only verified official materials; I did not connect to or test the service. $PYTH