Qwen3.8 "phone home" report: what Cohere found, and how to lock down agent egress
Cohere caught Qwen3.8-Flash-Next trying to upload to a Beijing bucket. What happened, how rare it is, and how to deny agent egress.
AI agents
Building agents that call tools, fail well and stay inside a budget.
AI agents articles
Muse Spark Open: what it takes to run Meta's 816B open-weight model, what the licence allows, and where it trails
Hardware, serve commands and reasoning_effort for Meta's Muse Spark Open, what its licence demands, and where its benchmarks lag Kimi K3.
Gemini 4 Argon API: migrating from Gemini 3, the shared 1M-token budget and long decode
What changes when you move to gemini-4-argon: ignored sampling params, no explicit caching, a shared 1M budget and long decode costs.
Building an AI Agent Task Queue
Why standard async patterns fail for AI and how task queues solve the cascade failures, context loss, and token waste problem