Meta is returning to the open-weight AI model category - a subset of large language models (LLMs) that it founded but then largely abandoned in a bout of misplaced priorities - with a loud and fairly sonorous bang, courtesy of the just-released Muse Glimmer open-weight model that employs creative tricks to make sure the model fits inside a single consumer GPU, and responds to queries in a lightning-fast manner. Meta used distillation to ensure that Muse Glimmer fits inside a single consumer GPU, while a tiny companion model speeds up response times Meta has just released the Muse Glimmer, a […]

Read full article at https://wccftech.com/meta-fires-back-at-chinas-15-week-ai-token-dominance-with-muse-glimmer-which-can-fit-inside-a-single-gpu-and-uses-an-innovative-technique-to-speed-up-responses/

It is wholly ironic that OpenAI, a company presumably founded on the principle that AI should be openly accessible to everyone, had refused to sign a letter advocating for open-weight AI models, with Sam Altman only conveying his support after the fact and under immense public pressure. At the same time, the CEO of Moonshot, a company that just unveiled the Kimi K3 open-weight AI model, one that seriously endangers the moat that both OpenAI and Anthropic were meticulously building around their proprietary models, has just issued a scathing take-down of OpenAI, demolishing its star credentials within the ecosystem. Moonshot […]

Read full article at https://wccftech.com/moonshot-ceo-says-openai-didnt-invent-anything-new-as-sam-altman-signs-the-letter-calling-for-open-weight-ai-models-after-the-fact-and-under-pressure/

Treasury Secretary Scott Bessent said the U.S. could sanction Chinese open AI models over alleged IP theft, expanding the Trump administration's campaign to slow China's AI advances.

Open source AI is booming, according to Hugging Face CEO Clem Delangue. The company has grown into something like a GitHub for AI in recent years, where AI builders can share and download open models and datasets, now used by roughly half the Fortune 500. Delangue has seen the same story play out again and again: companies start […]

New research suggests that AI memory systems can degrade model performance and encourage sycophantic tendencies.

If those same AI workloads can be handled by cheaper models without affecting quality, it would mean a massive shift in the economics of AI.

Mistral Forge lets enterprises train custom AI models from scratch on their own data, challenging rivals that rely on fine-tuning and retrieval-based approaches.