https://rundownsai.com/meta-gem-mfu-20-25-percent-kunlun-37/ Rundowns AI en 2026-10-18T22:25:18+00:00 Meta puts GEM training efficiency at 20-25%, its paper says 37% https://rundownsai.com/nvidia-nemotron-bolt-pinchbench-rank/ Rundowns AI en 2026-10-18T16:39:49+00:00 Nvidia bills Nemotron Bolt #1 on PinchBench, its card scores it #2 https://rundownsai.com/bank-of-america-mythos-remediation-unpriced/ Rundowns AI en 2026-10-18T00:51:37+00:00 Bank of America hired 1,000 contractors but never priced the Mythos work https://rundownsai.com/nvidia-european-city-ai-agents-10-percent-saving/ Rundowns AI en 2026-10-17T18:39:12+00:00 Nvidia names 8 European city AI projects, one with a 10% saving https://rundownsai.com/gemini-4-argon-safety-report-thought-penalty/ Rundowns AI en 2026-10-17T00:21:24+00:00 Google cut a thought penalty from Gemini 4 Argon’s early training