Kompletni AI platforma pro cesky financni a realitni trh. Microservices architektura, GPU inference, real-time data pipeline.
Distribuovana AI infrastruktura — 4 servery, 83+ PM2 sluzeb
Production microservices · Real-time · Enterprise-grade
31/31 modulu · 22-step cycle · Daemon mode · Auto-export 07:15
RTX PRO 6000 · Ollama inference · Production models
| Model | Params | Speed | VRAM | Use Case |
|---|---|---|---|---|
|
aurum-chat
Qwen3-235B
|
235B | 18 tok/s | 94 GB | Primary chat & reasoning |
|
aurum-brain
DeepSeek-R1:70b
|
70B | 32 tok/s | ~42 GB | Deep reasoning, analysis |
|
qwen3.5:35b
Qwen 3.5
|
35B | 139 tok/s ⚡ | 24 GB | Fastest · High-volume batch |
|
qwen2.5-coder:32b
Qwen 2.5 Coder
|
32B | — | ~20 GB | Code generation, Agent Zero |
6 major releases · Production-deployed · March 2026
Production · 83+ microservices · Czech market