Ant Group Launches Ling-3.2-flash, a 560B Native Multimodal Model 17x Cheaper Than Claude
Ant Group's Ling team launched Ling-3.2-flash on November 6, a native multimodal model with 560 billion total parameters, 25 billion activated and a 1-million-token context, Pandaily reported. Unlike bolt-on vision modules, it fuses visual and language at the architecture level and accepts image, text and video input. Its highlight is a visual-feedback self-check with adaptive error recovery: the model treats screens or interfaces as observations, compares structure and layout, and retries on failure. On DesignArena's image-to-frontend leaderboard it entered the top 10 with an Elo of 1,238. API pricing is $0.30 per million input and $0.90 per million output tokens - about one-seventeenth of Claude Opus 5 - and is free for two weeks; weights are expected to open soon.
Source: Pandaily / IT Home, Nov 6-7, 2026 | 2026-11-07
- China's Cyberspace Regulator Cracks Down on AI Impersonation of Public Figures
- Beijing E-Town Offers Up to 30 Million Yuan Subsidies for Super Agents and AI Terminals
- Agility Robotics IPO on Nasdaq Signals Global Humanoid Commercialization, Lifts China Suppliers
- CIIE Becomes AI Stage: Siemens Localizes R&D, Embodied-Brain Alliance and Open Physical-AI Platform Debut
- Nvidia NIM Opens Free Access to Chinese Open Models DeepSeek-V4.1-Flash, GLM-5.3-Flash, Kimi K3.1-Pro
- Moonshot AI Opens Kimi K3.1 Pro Weights: 2.8T Parameters, 1M Context, Plus Hosted Agents
- ByteDance's Doubao Work Deployed in Bosch Engineer Teams, Drafts 70-80% of Reports
- Yiwu Hosts Auto-Export Trade Show, Inaugurates Used-Car Base and 7-Nation Industry Alliance
- Yiwu Debuts 'Going-Global One Network' Service System at World Zhejiang Merchants Conference
- Yiwu Cross-Border E-Commerce Imports Hit 23.8 Billion Yuan, Builds 15-Minute Foreign Service Circle
