Mandarin Assistant - an inference router across China’s frontier AI models.
The problem is simple: there is no single best model anymore. DeepSeek, Kimi, Qwen, GLM and MiniMax all behave differently across coding, long-context, research and cost. Mandarin Assistant fixes it.
9 Actualites Decisives IA - Press Monitor
@UrsulaVonDerLeyen SamAltman Anthropic DarioAmodei OpenAI Siemens DeepSeek Qwen MarioDraghi EmergenceAI Claude MistralAI TotalEnergies Google AP_HP Thales DeepMind Opio Boulanger EmilieSoleri
If you're making 3D stuff with Astra or Deepseek or GLM via Blender MCP for threejs
You'd have noticed that trees are blobs and the branches are disconnected. I'm WIP on the home for my agents - anyway here's a prompt recipe on how to get better stylized trees below:
Same question, left and right comparison: Left: GPT6-Astra-High main Agent + DeepSeek-4.1-Flash sub-Agent + DeepSeek Harness
Right: GPT6-Astra-High + Codex To be honest, both sides are amazing. On the left, GPT6-Astra-High is responsible for the design, and DeepSeek-4.1-Flash is responsible for the implementation. This combination is already very capable, and the effect exceeds my expectations.
You have reached the end of the archive
All of deepseek