Tagged with

3 articles found

Qwen 3.8 Max Just Dethroned Opus 5 on the Agentic Index, For About Six Hours

Alibaba’s Qwen 3.8 Max briefly topped the Artificial Analysis agentic index. What the methodology update, benchmark drama, and open-weight release actually mean for AI teams.

#agentic AI#Artificial Analysis#open weights...

DFlash

Xiaomi’s 300B Model Just Got a Secret Speed Hack, DFlash Is the Real Deal

Xiaomi quietly dropped MiMo-V2.5-DFlash on Hugging Face. A 311B parameter model with block diffusion speculative decoding that could double your inference speed. Here’s what the community is finding.

#DFlash#LLM Inference#mimo...

abliteration

Abliteration Autopsy: 85 GPU-Hours of Forensics Reveal Which Safety Removal Actually Works

An open-source toolkit compared five abliteration methods on Qwen3.6-27B. The data exposes which techniques preserve capability, which destroy it, and why one popular method is built on stolen code.

#abliteration#LLM Safety#model alignment...