Model launchesPrismML releases 2-billion-parameter on-device LLM optimized for Snapdragon-based smart glassesPrismML, an AI lab founded by Caltech researchers and advised by UC Berkeley’s Ion Stoica, announced a 1-bit Bonsai language-vision model that can run locally on Qualcomm’s Snapdragon AR1 Gen 1 Platform.2 min read
Model launchesNVIDIA's NV-Reason-CT: a native 3D vision-language model for chain-of-thought CT analysisNVIDIA introduced NV-Reason-CT, a vision–language foundation model purpose-built to process full 3D CT volumes and produce structured reports with radiologist-style chain-of-thought reasoning and multiturn conversational follow-up.6 min read
Model launchesTypeSafe’s Jev: a decision-first model that’s much cheaper and faster but offers no explanationsTypeSafe has released Jev, a model designed only to make calibrated decisions—returning typed answers with probabilities instead of generating free-form text.5 min read
Model launchesGoogle debuts Gemini 3.8 Flash and Flash‑Lite text‑to‑speech modelsGoogle introduced two new Gemini text‑to‑speech models, gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts, offering a library of over 2,000 voices and the option to create a custom voice from a 30‑second sample.2 min read
Model launchesFalling prices for powerful AI models drive usage growth and bolster AI boom outlookMajor AI developers are cutting operational costs for their top models, leading to sharply higher usage.3 min read
Model launchesQualcomm unveils Snapdragon 8 Elite Gen 6 chips focused on on-device AI and pro videoQualcomm introduced two flagship SoCs, the Snapdragon 8 Elite Gen 6 and Snapdragon 8 Elite Extreme Gen 6, emphasizing on-device AI, personalization and professional video capabilities.3 min read
Model launchesAnthropic launches Opus 5.5 with faster performance and lower pricesAnthropic released Opus 5.5 on Tuesday, a faster and cheaper update to its top-tier Claude model that the company says outperforms larger models like Fable on many benchmarks.2 min read
Model launchesOpenAI forms independent math advisory group after rapid internal AI breakthroughsOpenAI says an internal model trained starting August 28 has solved the Navier–Stokes Millennium Prize problem and more than 100 long‑standing mathematical problems, prompting internal debate about communicating such progress.3 min read
Model launchesOpenAI expands capabilities as investors pour billions and researchers raise safety alarmsOpenAI announced new model capabilities for software control, scientific reasoning and financial services while reporting a coordinated AI effort on the Navier–Stokes problem.3 min read
Model launchesRunning Prism's Ternary Bonsai 2 27B GGUF with Prism's llama.cpp forkA user reports steps to run Prism’s compressed Ternary-Bonsai-2-27B GGUF model locally using Prism’s fork of llama.cpp and its bundled llama-server.3 min read
Model launchesPrismML shrinks LLMs to run locally — Bonsai 2 27B compresses Qwen3.8 to 5.9 GBPrismML, a Caltech-founded startup led by Babak Hassibi, released Bonsai 2 27B, a compressed version of Alibaba’s Qwen3.8 27B that fits into 5.9 GB of storage — a roughly 9–10× reduction in memory with about 98% of the original’s benchmark performance.3 min read
Model launchesOpenAI launches 'Astra for Law'—GPT‑6 Astra tailored for law firms and legal techOpenAI introduced Astra for Law, a specialized legal foundation built on GPT‑6 Astra that combines a legal search index, custom instructions, and integrations for law firms and legal‑tech companies.5 min read
