Open source token efficiency benchmark update, inference taxonomy framework, Alec Aivazis on Houdini and GraphQL

An updated open source benchmark suite shows GraphQL is more token-efficient than REST and other JSON API technologies for connecting agents to deployed services, according to the benchmark's new numbers. The same update presents a framework for categorizing inference calls to identify "code smells" that waste money on low-value tokens, and features a conversation with Alec Aivazis, creator of the Houdini GraphQL framework.

Benchmark: GraphQL is more token-efficient than REST benchmark-graphql-is-more-token-efficient-than-rest New numbers from my open source benchmark suite show a significant and durable advantage for using GraphQL rather than other JSON API technologies under the hood to connect agents to deployed services. Categorizing and evaluating inference spend categorizing-and-evaluating-inference-spend Not all tokens give equal value. Here's a framework for categorizing your inference calls and finding "code smells" that might be costing you a ton of money for relatively little value. Alec Aivazis, creator of Houdini GraphQL alec-aivazis-creator-of-houdini-graphql Alec is a consummate GraphQL practitioner and tooling visionary. The Houdini framework is a unique take on the GraphQL developer experience and it was an absolute pleasure hearing Alec's take on this and much more in these conversations: