Qwen3.6-35B-A3B

Goodfire argues individual neurons can explain how LLMs compute
Goodfire Research says individual neurons in a large language model can serve as meaningful units for understanding computation, challenging the view that researchers…

PyTorch 2.15 brings FP8 and much faster MoE decoding to Macs
PyTorch 2.15 adds float8_e4m3fn support on the MPS backend and, according to the Hugging Face PyTorch team, makes mixture-of-experts models decode 5 to 6.5 times faster…

Liquid AI releases LFM2.5-24B-A2B, an open-weight model for on-device agents
Liquid AI has released LFM2.5-24B-A2B, a 24B-parameter reasoning model with only 2B active parameters, designed for fast, private agents that run on-device. According to…

Modal shows stable RL training with mixed-precision, mixed-accelerator samplers
Modal says it has trained a large model with reinforcement learning (RL) using samplers on six kinds of accelerators and several numeric precisions, matching a…