LongCat-Flash-Lite-Sparse
Meituan · China
Meituan long-context MoE model
- Release month
- Field
- Chat
- Availability
- open-weight
- Parameters
- 69B-A3B
- License
- mit · Commercial OK
Overview
- Registered on Hugging Face July 31, 2026 · Open weights · 69B (about 3B active)
- Sparse attention for more efficient long-input processing · Non-thinking · Up to 1M context · MIT
View official announcement ↗
← All models