Liquid AI, released 2026-05-28. Dense hybrid decoder, 1.2B parameters, 128K context. Attention: LIV convolution blocks plus GQA.
Dense hybridLiquid AI 🇺🇸 · May 28, 2026
LFM2.5 (350M)
Liquid AI's compact LFM2.5 model combines double-gated LIV convolution blocks with periodic GQA layers.
Key detail
Uses a 16-layer hybrid stack with 10 LIV convolution layers and 6 GQA layers.Parameters
350M
Context
128K
KV / token
12 KiB
Fact sheet
- Scale
- 350M parameters
- Context window
- 128,000 tokens
- Decoder type
- Dense hybrid
- Attention
- LIV convolution blocks plus GQA
- Layer mix
- 10 LIV conv + 6 GQA
- KV cache / token (bf16)
- 12 KiB · Very low
- License
- LFM Open License v1.0
- Released
Efficiency
Parameters active per token
100%
Dense: every weight runs for every token.
Architecture concepts
Sources & resources
Use the ← and → keys to step through releases.
Similar architectures
Liquid AI, released 2026-10-27. Dense hybrid decoder, 1.7B parameters, 256K context. Attention: ShortConv plus gated GQA with QK-Norm and sliding-window/global attention.
NVIDIA, released 2026-03-16. Dense hybrid decoder, 4B parameters, 256K context. Attention: GQA with only 4 attention layers.
Alibaba Cloud, released 2025-04-28. Dense decoder, 600M parameters, 32K context. Attention: GQA.