Executive Summary
Ant Group has launched Ling 3.0 Flash, a 124B parameter Mixture-of-Experts (MoE) model, now available on the AI Gateway platform. The model is designed for token-efficient, production-scale agentic workflows, featuring a large 256K token context window. It is targeted at developers building applications for coding, document analysis, and long-context interactions, and is being offered for free for a three-week promotional period.
Key Takeaways
* Model Name: Ling 3.0 Flash
* Architecture: Mixture-of-Experts (MoE) with 124B total parameters and approximately 5.1B active parameters per token.
* Context Window: 256K tokens.
* Primary Use Case: Optimized for high-frequency, multi-step agentic workflows, coding agents, document processing, and long-context interactions within tight token, latency, and cost budgets.
* Availability: Accessible now on AI Gateway using the model identifier `inclusionai/ling-3.0-flash-free`.
* Promotional Pricing: Free to use for three weeks, through August 3rd.
Strategic Importance
This launch provides developers with a powerful, cost-effective model specialized for building complex AI agents, increasing competition in the API-accessible model market.