https://bitbytecore.com/article/gemini-flash-pricing-has-an-expiry-date BitByteCore en 2026-09-03T09:32:00.304Z Gemini Flash is half price until 31 December, and Google published the date it ends https://bitbytecore.com/article/what-ai-assistants-actually-cite-when-you-ask-what-software-to-buy BitByteCore en 2026-09-03T07:05:02.350Z Asked which software to buy, an AI assistant cited a demo vendor's blog more often than Gartner https://bitbytecore.com/article/on-device-ai-on-phones-privacy-and-latency-not-a-gimmick BitByteCore en 2026-09-03T05:56:00.000Z On-Device AI on Phones: Privacy and Latency, Not Hype https://bitbytecore.com/article/ai-agents-are-moving-from-demos-to-narrow-jobs BitByteCore en 2026-09-03T05:55:00.000Z AI Agents Are Moving From Demos to Narrow Jobs https://bitbytecore.com/article/mistral-trains-on-your-chats-by-default-how-to-opt-out BitByteCore en 2026-09-02T20:13:05.538Z Mistral trains on your chats by default. Turning it off takes three separate switches https://bitbytecore.com/article/gigapath-flash-distilled-pathology-foundation-models BitByteCore en 2026-09-02T13:16:21.315Z Microsoft distilled a billion-parameter pathology model into 22M parameters https://bitbytecore.com/article/gemini-agentic-video-understanding-token-savings BitByteCore en 2026-09-02T10:39:43.853Z Gemini can now skim a video instead of watching every frame https://bitbytecore.com/article/world-labs-atlas-world-model-benchmarks BitByteCore en 2026-09-02T10:31:53.926Z World Labs' Atlas wins six of seven 3D benchmarks, and you cannot run it https://bitbytecore.com/article/claude-fable-5-1-cheaper-without-a-price-cut BitByteCore en 2026-09-02T07:15:34.968Z Claude Fable 5.1 costs 25% less, and its per-token price did not move https://bitbytecore.com/article/small-models-are-eating-the-easy-work BitByteCore en 2026-09-02T05:56:00.000Z Small Models Are Quietly Taking Over the Easy Work https://bitbytecore.com/article/which-ai-subscription-should-you-actually-pay-for BitByteCore en 2026-09-02T05:55:00.000Z Which AI subscription should you actually pay for? https://bitbytecore.com/article/deepseek-v4-vision-model-open-weights-307gb BitByteCore en 2026-09-01T17:05:21.676Z DeepSeek's first V4 vision model is MIT-licensed, and 307 GB https://bitbytecore.com/article/gemini-flash-open-problems-antigravity-teamwork BitByteCore en 2026-09-01T16:02:27.144Z A cheap model solved open math problems, once Google put a team around it