Revenue shortfall shakes investor confidence OpenAI's annualized revenue is now estimated at $50 billion, $20 billion below the $70 billion figure reported earlier. The revision comes from a TechCrunch analysis that c…

Math output quality falls short of community standards OpenAI's recent wave of mathematical proofs failed to follow guidelines set by a consulted group of researchers. The discrepancy was highlighted in a TechCrunch r…

Community‑driven Lemonade brings inference home Lemonade launched as an open‑source server that runs large language models on a user's GPU or NPU. The project claims parity with cloud APIs while keeping data on the lo…

Technical trade‑offs of on‑device inference Running models locally eliminates network latency, but it demands substantial compute resources. A high‑end GPU can handle a 7 billion‑parameter model at acceptable throughp…