Ant's Ling-3.2-flash: 560B/25B-active multimodal model demoed on PDFs, Blender, agent office work
Fancy-Equivalent-284 · reddit · 2026-11-06
Ant Group's Ling team launches Ling-3.2-flash, a natively multimodal model with 560B total parameters (25B active) supporting text, image and video input, focused on visual understanding and task execution.
Four concrete demoed workflows:
- Reading a research PDF's figures and equations, then implementing the method in working code
- Operating Blender through interface actions to build a furnished room
- Recreating 3D assets from 2D reference images
- Coordinating specialist agents to turn Q2 financial data into an analysis report, a six-slide board deck and a highlights video
The launch includes an office-work demo showing the generated documents and video. The model is available via the documented APIs and listed on Computrix.
More from coding & agent
- Ground Truth MCP connector keeps AI agent answers fresh with monitoring and alerts — modelcontextprotocol · 2026-11-06
- Reddit MCP Server lets AI agents read and interact with Reddit via OAuth2 — modelcontextprotocol · 2026-11-06
- playmcp turns any track into a playable music game via AI agents — WhaleSong2077 · 2026-11-06
- Supabase November update: Grok Bot integration lets teammates request changes in chat via Cursor — dshukertjr · 2026-11-06
- Open-source Apiarius adds a live deck for watching Claude Code subagents in real time — kontrollable · 2026-11-06
- Frontier3DBench: 145-task benchmark tests coding agents on building interactive 3D worlds — _akhaliq · 2026-11-06