Meituan Open-Sources LongCat-Flash-Lite-Sparse: 256k Context on a 24GB GPU
Gohab2001 · reddit · 2026-07-31
Meituan has released LongCat-Flash-Lite-Sparse, an open-source MoE model. It features roughly 3B active parameters and offloads a 30B n-gram lookup table to RAM.
This architecture allows the model to support a fast 256k context length using only a 24GB GPU. The reviewer notes that this technique is reminiscent of Gemma 4's PLE trick, though initial analysis suggests it won't replace heavier models like Qwen 3.6 27b in overall performance.
Related event: Meituan Open-Sources LongCat-Flash-Lite-Sparse Model(4 posts)→
More from Models
- Trying Grok bot surfaces an unpaid Cursor balance, blocking account merging — haydendevs · 2026-10-16
- TensorFold 1.0.12 runs 120B vision model on 1-2 DGX Sparks, adds Mac image input — HankYeomans · 2026-10-16
- Leaked screenshot shows unannounced Gemini 4 Argon gated to Ultra and paid Pro subscribers — il_-bro · 2026-10-16
- Google: Gemini 4 Argon lands in Antigravity, code submissions up 35%, rollbacks down ~33% — tekbog · 2026-10-16
- Google's "ungrounded beliefs" classifier injections derail Opus 4.1's reasoning chain — liminal_bardo · 2026-10-16
- A week of ChatGPT Pro quota burned in 36 hours as GPT-6 fails to factor a 512-bit RSA key — matthew_d_green · 2026-10-16