Skip to content
AI News HubLIVE
In-site rewrite2 min read

Parse, Extract, Classify — now each with its own MCP | Weekly Newsletter

Summary

LlamaIndex team doubles in size, releases updates: conversational extract, markdown in fast tier, improved tables in Agentic Plus, and 100 users on every plan. AI news includes GPT-5.6 benchmark, Bun rewriting Zig in Rust, Apple suing OpenAI.

Parse, Extract, Classify — now each with its own MCP | Weekly Newsletter
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

Hi there, LlamaParser!

Quite the week. Our team at LlamaIndex has doubled in the past year, so we met up at HQ in SF to plan ahead and ship new features now available: Extract now has conversation, fast tier now includes markdown, Agentic Plus got noticeably better at tables, and every plan now seats 100 users. Here's the rundown...

Product Updates

Writing JSON Schema by hand? That's over.⁣ Your extraction schema is now a conversation away. See Conversational Extract inaction →

Fast tier now outputs markdown by default — set tier="fast", version="latest" and you get clean markdown in the API and the UI. It's LiteParse under the hood, with better OCR and better PDF font handling. Try it in the playground →

100 users on every plan — including free. Bring your whole team.

Parse playground polish — cleaner settings, easier mode selection, and a quick confirmation before you accidentally parse a 400-page PDF you only wanted to peek at.

Darkmode — give your eyes a break and try the new darkmode UI

Community Updates

Swap and Pitch — Thursday, July 16, San Francisco — demos and pitches at 855 Brannan. Come say hi. Register →

AWS AI Builder Lab — Friday, July 24, New York — a hands-on day of agent building at Washington Square Park South. Register →

You.com Agentic Hackathon — Friday, July 24, San Francisco — build agents with us at the AWS Builder Loft, 525 Market St.

AI News

GPT-5.6 hits ParseBench on day 0 — text and tables: strong. Charts and layout: still a struggle. Most interesting: Luna is ~6x cheaper than Sol with only minor degradation across every ParseBench.ai metric — more reasoning tokens don't buy you more visual understanding.

Bun rewrote 535k lines of Zig in Rust — in 11 days, with 64 Claudes, for $165k in tokens — and the Zig creator's rebuttal is the sharpest thing you'll read this week. The rewrite → · The rebuttal →

Apple is suing OpenAI over trade secrets — from ChatGPT-on-iPhone partners to federal court in under two years. Read more →

TypeScript 7 is GA, Go compiler and all — the ~10x compile-speed rewrite finally lands. Watch →

Chart of the day: US IPO Market on course to reach new records fueled by AI

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • LlamaIndex product updates: conversational extract, markdown output in fast tier, Agentic Plus table improvements, and 100 users per plan.
  • Community events in San Francisco and New York.
  • AI news: GPT-5.6 on ParseBench, Bun's Zig-to-Rust rewrite, Apple suing OpenAI.

Highlights and analysis are generated automatically and may contain errors. Check the original source.