https://dangmua.com/blog/a-new-tokenizer-cut-one-teams-inference-cost-by-77 DangMua en 2026-10-21T18:23:38.000Z A New Tokenizer Cut One Team’s Inference Cost by 77% https://dangmua.com/blog/openai-agent-disabled-its-safety-monitor-then-denied-it DangMua en 2026-10-21T18:17:59.000Z OpenAI Agent Disabled Its Safety Monitor, Then Denied It https://dangmua.com/blog/gemini-personal-intelligence-hits-android-auto-what-it-reads DangMua en 2026-10-21T11:21:34.000Z Gemini Personal Intelligence Hits Android Auto: What It Reads https://dangmua.com/blog/whose-settings-does-your-ai-agent-obey-hostile-repo-test DangMua en 2026-10-20T18:18:49.000Z Hostile Repo Test: Whose Settings Does Your Agent Obey? https://dangmua.com/blog/anylanguagemodel-1-0-which-local-model-fits-your-device DangMua en 2026-10-20T18:16:34.000Z AnyLanguageModel 1.0: Which Local Model Fits Your Device https://dangmua.com/blog/mercury-3-hits-1-068-tokens-second-what-to-test-first DangMua en 2026-10-20T11:17:10.000Z Mercury 3 Hits 1,068 Tokens/Second: What to Test First