DeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that’s a nightmare for OpenAI
DeepSeek's free 685B-parameter AI model runs at 20 tokens/second on Apple's Mac Studio, outperforming Claude Sonnet while using just 200 watts, challenging OpenAI's cloud-dependent business model.
A new, challenging AGI test stumps most AI models
The Arc Prize Foundation has a new test for AGI that leading AI models from Anthropic, Google, and DeepSeek score poorly on.
DeepSeek releases improved V3 model under MIT license
DeepSeek today released an improved version of its DeepSeek-V3 large language model under a new open-source license. Software ...
Tencent upgrades AI model to compete with DeepSeek, Alibaba
Tencent Holdings launched an upgraded AI model, Hunyuan T1, to enhance its competitive edge in China’s AI market. T1 matches ...
The Business Times18h
AI chip startup FuriosaAI rejects Meta’s US$800 million offer
SEOUL] Korean chip startup FuriosaAI has turned down an US$800 million takeover offer from Meta Platforms, choosing instead to grow the business as an independent company, a person with knowledge of ...
New York Magazine18h
Nick Denton Has Had Enough of America
The former media mogul discusses how he became “ashamed” of Gawker and why he’s selling his apartment and moving to Budapest.
Big Tech rallies behind US President Trump’s proposal to ease regulations around AI
Trump’s light-touch approach to regulating AI runs counter to the efforts of the EU, which introduced its landmark AI Act ...