Unverified50% confidenceReleaseExact time
Ant has released the Link 3.1 Flash weights, using a hybrid architecture of linear and full attention; the model card lists 560 billion total parameters and 25 billion activated parameters, with the license marked as MIT.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/31/2026
First Seen
Valid until: 1/29/2027
Sources
Related Entities
Related Claims
UnverifiedQwen3.8 Flash Next additionally introduces an Ngram module with 51 billion parameters, resembling a massive local phrase memory table.70% similarUnverifiedv4.1 Flash introduces a mechanism called Ngram, where 2 out of 40 layers contain conditional memory tables, with 196 billion parameters used for a lookup table that matches sequences of up to 4 tokens against a vocabulary of 16 million.69% similarUnverifiedAnt Group's Bailing open-sourced the lightweight model Ling 3.0 Tiny on the Hugging Face platform, adopting a hybrid reasoning MoE architecture with 7.9B total parameters and only 1.3B activated per token.68% similarUnverifiedQwen3.8 Flash Next is a Flash model using the next-generation Qwen4 architecture, and the complete weights were released on the day of its open-sourcing.64% similarUnverifiedLing-3.1-flash adopts a Mixture of Experts (MoE) architecture with a total parameter count of approximately 560 billion (~560B).64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1119118API
curl https://kongchang.com/api/v1/knowledge/claims/1119118MCP
get_claim(id=1119118)