Hugging Face publishes Phonon text-to-speech retrospective
Hugging Face published a retrospective on Phonon, its small on-device text-to-speech model: two sizes, licenses, and what runs free.
Based on: Hugging Face
Models
Major AI models and their updates: what they can do and why it matters for content work.
-
-
Mistral Large 4 (Le Chonk) open weights published on Hugging Face
Mistral released a 1T-parameter open-weight multimodal model on Hugging Face, with a 1M-token context window and limits on model-as-a-service use.
Based on: Hugging Face
-
Google DeepMind announces Gemini 4 Carbon, a cybersecurity-focused model
Google DeepMind announced Gemini 4 Carbon, its next frontier model focused on cybersecurity. Access starts with cyber defenders only.
Based on: Google DeepMind
-
Underdog shows Woof, a 32MB voice model, in a browser game
Underdog released a browser game to demo Woof, a 32MB speech model with 31 million parameters that the company says runs on most devices without internet.
Based on: TechCrunch
We retell the news in our own words and always link to the original source. Texts are drafted by AI and checked against the source. RSS feed