Meanwhile on my free models: glm flash and deepseek instant
Built custom pluggin directly inside Hermes Desktop!
🛠️ Got my spare rig running Omarchy OS with Hermes orchestrating live routing between DeepSeek-V4-Flash (on the DGX Spark for reasoning) and Qwen3-VL (on the RTX 3060 for vision). Building local AI just got way more fun. ⚡
DeepSeek heavy investment in computing center
Verdent AI just became much more useful for actually building things.
The smartest part is knowing which model to use for which job. Here’s the workflow: → GLM 5.3 Use it for complex coding, debugging, code review, and harder reasoning. → GLM 5.3 Flash Use it for quick
You have reached the end of the archive
All of deepseek