Part 3: `tokenprice` ($0.001) vs the price-feed incumbents — measured, not marketed

Part 3 of the x402 benchmark series examines the differences between the tokenprice endpoint and the incumbent price feeds, CoinGecko and CoinMarketCap, in terms of cost and performance. The tokenprice endpoint charges $0.001 per call, significantly lower than the $0.01 per call charged by CMC's pay-per-call model. This lower cost makes tokenprice more attractive for experimentation and development of trading agents, portfolio trackers, and research bots.

The benchmark measures the latency and data provided by each price feed. The tokenprice endpoint demonstrates a median latency of 1.8 seconds for various queries, such as WETH by address or USDC by address. Despite this fast performance, the endpoint makes a crucial distinction between querying by address and ticker symbol, as ticker queries can be ambiguous and may return the wrong token.

While the per-call pricing model proves cheaper for low-volume use cases (under 30,000 calls per month), the subscription model of CoinGecko and CoinMarketCap becomes more cost-effective at higher call volumes. The tokenprice endpoint emphasizes its focus on providing live quotes and DEX liquidity, without offering extensive historical data or coverage of an extensive number of coins.

To try out the tokenprice endpoint, readers are provided with a free trial via a curl command, allowing them to test the service without signing up. The pricing information for all 43 endpoints is also available on the API pricing page. As the series continues, the next part will compare websearch, Exa, and Tavily in head-to-head competition.