LlamaIndex CEO Jerry Liu: 'The framework era is over.' Pivoting from open-source scaffolding to managed doc infra via LlamaParse and LlamaCloud — context quality as the new moat.
ParseBench at CVPR 2026: the first AI-agent document benchmark — 2K+ verified pages, 167K+ test rules, 5 dimensions. ArXiv 2604.08538. Open source on HuggingFace and GitHub.
LiteParse v2 is a Rust rewrite of LlamaIndex's PDF parser achieving 100× speed gains, sub-second on 450-page docs, with WASM and multi-language bindings now live.
Google I/O 2026 deprecated Gemini CLI in favor of Antigravity, launched Gemini for Science with 100+ institutions, and unveiled native video editing via Gemini Omni Flash.
LlamaIndex launched LlamaParse Mobile — an iOS/Android app that converts a phone camera photo into clean, copyable document text in under one minute.
LlamaIndex named to CBInsights AI 100 2026 (AI Infrastructure category) — positioned as 'the leading document understanding API for AI agents' targeting the world's PDF data layer.
LlamaIndex releases a production pipeline: LlamaParse extracts from tax returns, pay stubs, W-2s, and bank statements; Claude validates consistency, catches gaps, and outputs a COMPLETE/REVIEW/FLAG decision.
LlamaIndex's ParseBench launches on Kaggle: 2K enterprise pages, 167K+ test rules, 5 dimensions, with Gemini 3 Flash leading the current board.
LlamaIndex open-sources LiteParse—a VLM-free, ML-free PDF parser using a 6-step grid projection algorithm that handles tables and complex layouts fast.
Curated AI insights — sent when there's something worth your inbox.