United Imaging Intelligence has open sourced a medical video AI model along with a massive dataset and benchmark, showing how smaller specialized models can outperform general purpose AI in clinical tasks.
Mozilla’s MZLA Technologies has introduced Thunderbolt, an open-source AI client designed to give organizations more control over how AI runs inside their infrastructure. The project focuses on self-hosting, flexible model choices, and enterprise integrations through the Haystack AI framework.
IBM says artificial intelligence is moving toward infrastructure status and argues that open source development may be the safest way to build and secure it.
A newly released open source framework called SentiAvatar aims to make AI avatars behave more like real people during conversation. The project includes a motion dataset with 21,000 clips, a foundation model trained on hundreds of hours of motion data, and a system that synchronizes speech, facial expressions, and body gestures in real time. If the technology works as advertised, it could help developers build more believable NPCs, AI assistants, and virtual characters.
Google and the Kubernetes community have launched a new AI Conformance program designed to ensure Kubernetes clusters can properly support machine learning workloads, GPUs, and large scale inference.
Mozilla is partnering with Mila to push open source AI forward, focusing on privacy, trust, and giving developers and governments more control.
South Korea is making a serious play for AI independence. SK Telecom’s 519B parameter A.X K1 model is not about consumer hype, but about building national AI infrastructure outside US and Chinese control.
Z.ai is not a well known name, but its open source GLM 4.7 model targets a real developer problem: keeping AI stable across long running coding workflows.
OpenAI just released gpt-oss-120b and gpt-oss-20b, two powerful open-weight AI models optimized for reasoning, tool use, and on-device deployment.