Perplexity's post-trained GLM 5.2 preview pairs a cheaper open-source base with a frontier advisor—and it's already live in ...

General-purpose models struggle with messy, industry-specific data. A three-layer AI stack from Trunk Tools cut document ...

We fine-tuned on AMI, the most gamed testset on the HF Open ASR Leaderboard, and climbed 10 places. Here is what benchmaxxing ...

The MiniCPM5 cognitive core brings 1 billion parameter AI to smartphones and older hardware, enabling offline reasoning and ...

In enterprise AI, the models that win won't be the largest. They'll be the ones that know exactly what they're built for.​​ ...

Indian AI startups, have been using open-weight models to build enterprise AI applications for some time. Mint explains why.

The rise of AI has brought an avalanche of new terms and slang. Here is a glossary with definitions of some of the most ...

Local AI inference at 32B-parameter quality, no cloud API required: University of Waterloo researchers released PAW on July 2 ...

LLM training data mixture optimization breaks when training pools shift — every prior proxy experiment becomes stale.

Local AI inference at 32B-parameter quality, no cloud API required: University of Waterloo researchers released PAW on July 2 ...

Thomson Reuters’ AI approach shows why trusted outputs depend on authoritative data, rigorous governance, and domain ...

Tencent's Hy3 is a 295B mixture-of-experts model that runs on 21B active parameters, closes the gap with closed frontier labs ...