LlamaIndex 发布 ParseBench 基准及本地解析工具更新

Hi there, Llama Enthusiasts!

Welcome to this week's edition of the LlamaIndex newsletter! We're excited to share major updates including our new document parsing tools for local and private deployments, the introduction of sandboxed-lit for secure agent interactions, and insights from our recent events in Singapore and NYC. Plus, we're unveiling ParseBench, the first OCR benchmark designed specifically for AI agents.

If you haven't explored LlamaParse yet, make sure to sign up and get in touch with us to discuss your specific enterprise use case.

🎯 ParseBench Webinar Alert

• Join our live webinar to discover ParseBench, the first document OCR benchmark built specifically for AI agents. Learn how current benchmarks fall short of what AI agents actually need and how to evaluate the accuracy of your document ingestion pipelines. Register now

🤩 Open Source Updates

• Sandboxed-Lit CLI Agent: A Rust-powered CLI agent that combines LiteParse document parsing with secure sandbox environments, allowing agents to safely interact with PDFs, images, and Office documents while maintaining full filesystem access. GitHub, Learn more about LiteParse
• LiteParse-Server: Self-hostable, open-source HTTP server for private document parsing that stays 100% local. Deploy as a Docker container or serverless Express.js API with Redis caching, OpenTelemetry integration, and production-ready observability. Blog post, GitHub repo

• Singapore Wrap-Up: Successfully concluded a live workshop and keynote at AI Engineer Singapore. Looking forward to seeing everyone at the upcoming world fair in SF!
• NYC Developer Events: Hosted back-to-back events including a hands-on developer workshop led by Jerry Liu and Logan Markewich, plus an AI engineer happy hour, both reaching capacity with strong community engagement. For early access to future events, subscribe to our Luma.