LlamaIndex Introduces Turbo Mode for Extract

LlamaIndex posted on X that it has added Turbo mode to Extract. The company states the new option pulls structured data from documents roughly four times faster than its Cost Effective Tier while keeping comparable accuracy. It reports a median latency of 3.7 seconds per page. The post notes that pages run in parallel so overall latency stays near constant. Jerry Liu, co-founder and CEO, retweeted the announcement. An attached scatter plot chart shows latency data for the feature.