The open‑source Llama inference library published version b11757 with a fix for CUDA graph handling and new binaries for macOS, iOS and Linux.
NoteBy OossaPublished by Oossa: 1 min read
The ggml‑org team released llama.cpp version b11757. The update clears CUDA graphs after a legacy memory‑pool flush, which should reduce GPU memory leaks. Pre‑built binaries are now available for Apple Silicon macOS, Intel macOS, iOS, and several Ubuntu builds (CPU, Vulkan and CUDA 12.8). The release is posted on GitHub.
Developers running Llama models on Nvidia GPUs will see fewer out‑of‑memory errors after the fix.
Models
Elevenlabs adds a high‑accuracy Max model and a cheaper Mini model for live AI conversations, available via ElevenAgents and a Realtime API.
Models
The AI companion hits 10.2 M U.S. installs, beats ChatGPT’s early numbers and shows higher short‑term retention.
Models
Amazon Bedrock now lets developers in Canada call the latest OpenAI GPT‑5.6, GPT‑6 and GPT‑6.1 models via cross‑Region inference.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
