Liquid AI released LFM 2.5 QAD, a quantized version of its Liquid Foundation Model, with a 2.6B parameter GGUF file available on Hugging Face. The release was announced via a tweet from Liquid AI's official account.
nex-agi releases Nex-N2.5-mini, a 35B model, on Hugging Face
nex-agi releases Nex-N2.5-mini, a 35B model, on Hugging Face
LiquidAI released two new decision models, d1-3B and d1-omni-600M, with weights and GGUF quantizations on Hugging Face. The drop followed same-day community anticipation of a new LFM release teased by the company. No new LFM shipped, disappointing some community members.
LiquidAI released two new decision models, d1-3B and d1-omni-600M, with weights and GGUF quantizations on Hugging Face. The drop followed same-day community anticipation of a new LFM release teased by the company. No new LFM shipped, disappointing some community members.
A blog post on sander.ai introducing Continuous Diffusion Language Models (CDLMs) was shared on Hacker News, drawing 45 points of engagement. The post examines diffusion-based language modeling with continuous representations. It offers practitioners a research perspective on alternatives to autoregressive token prediction.
A blog post on sander.ai introducing Continuous Diffusion Language Models (CDLMs) was shared on Hacker News, drawing 45 points of engagement. The post examines diffusion-based language modeling with continuous representations. It offers practitioners a research perspective on alternatives to autoregressive token prediction.
Unsloth is updating its Qwen3.8-Flash-Next-GGUF repository on Hugging Face to work with llama.cpp. Community members hope this resolves the fragmentation of two different GGUF versions of Qwen Flash Next circulating locally.
Unsloth is updating its Qwen3.8-Flash-Next-GGUF repository on Hugging Face to work with llama.cpp. Community members hope this resolves the fragmentation of two different GGUF versions of Qwen Flash Next circulating locally.
Community claims Qwen 3.8 follows GPT-5.5 Pro reasoning prefills
Community claims Qwen 3.8 follows GPT-5.5 Pro reasoning prefills