A model whose trained parameters, its weights, are published so anyone can download, run and modify it on their own hardware.
Open weights are not the same as open source: the training data and code usually stay private, and the licence may still limit use.
Original games and genre games remain allowed, but the limit may only deter low-effort copycats, since users can turn to open-weight models without such rules.
Organisations that run the weights themselves keep access even if a hosted provider restricts or cuts it off.
The episode has renewed questions about whether restrictions on Chinese AI should cover open-weight models run on government machines.
The open-weight model runs on a laptop and needs more guidance than large paid coding assistants.
The chip startup stays independent and uses the new funding to scale hardware for open-weight AI models and video generation.
Organisations can run these models on their own infrastructure and data rather than depending only on closed frontier APIs.
It gives developers one embedding model for multiple media types, with smaller and larger configurations for different device limits.
Meta says the model can run on Apple’s M5 Ultra Mac Studio, but only with enough memory to spread it across two 512GB machines or four 256GB systems.
The update gives teams a longer-running agent workflow while adding tighter controls for AI usage and billing.
The 9B and 27B models are available in Workers AI and on Hugging Face, with a faster Flash variant for lower-latency decisions.
The model is meant to let platform teams update moderation rules without retraining, so policy changes can be reflected faster.
The findings add to concerns that open-weight models can be pushed past their safety layers and used for abuse.
The voice AI provider’s mix of big enterprises and creators shows how synthetic agents are moving into mainstream customer support and media work.
Open‑weight medical AI models are moving into real‑world screening and triage, from rural clinics to national health programs.