LiquidAI Releases GGUF Version of LFM2.5 1.2B with Speculative Decoding

LiquidAI · hf · 2026-08-19

LiquidAI has released a GGUF version of the LFM2.5-1.2B-Instruct-DSpark model. Based on the LFM2.5 architecture with 1.2B parameters, it is fine-tuned for text generation tasks. Key features include support for speculative decoding and functioning as a draft model, aimed at enhancing inference efficiency via llama.cpp.

Related event: LiquidAI's LFM2.5-DSpark Gets 3.2x Speedup with Speculative Decoding, GGUF Released(2 posts)→

Original post →

More from Models

Models channel →