Nov 8, 2026On-premise AI Lab🧪 Official multimodal 290k downloads, 9B uncensored, zero GPU voice: three offline tools this episode | Local AI LabDeepSeek's official new multimodal "Lightning" version hit 295,000 downloads in a single week, with the community cooking up three offline power tools: a 9B uncensored model, a pocket-sized TTS, and more; this episode skips cloud APIs to round up practical "download and run locally" options, which one fits a 16GB GPU for screenshot-to-JSON, which lets a laptop code and write novels offline, and where to find the bilingual speech model that runs without a GPU.
Nov 7, 2026AISelf-Host a Local AI API Gateway with LiteLLM: One Entrance to Manage All Models, Handling Usage, Costs, Failover, and API Keys in One GoSelf-host an on-prem AI API Gateway with LiteLLM Proxy: a single OpenAI-compatible endpoint integrating 29 cloud and local models, with use-case aliases, usage and cost dashboards, multi-tier Fallback chains, and virtual key management, plus a complete config.yaml example.
