Build Configurations

Recommended server configurations for different budgets and workloads.

---

Budget Build: €1,500-2,000

Target: Small businesses, experimentation, light inference workloads

Component	Spec	Est. Price
GPU	2x Tesla P100 16GB	€600-800
Server Chassis	Used Dell/HP/Supermicro 2U	€300-500
CPU	Xeon E5-26xx (included)	—
RAM	64GB DDR4 ECC	€150-200
Storage	1TB NVMe SSD	€100-150
Total		€1,150-1,650

What it runs:

• LLaMA 2 7B, 13B (full precision)
• Mistral 7B
• Code models (CodeLlama 7B/13B)
• Stable Diffusion (slow but works)

---

Value Build: €3,000-4,000

Target: Production inference, multiple concurrent users

Component	Spec	Est. Price
GPU	2x Tesla V100 16GB	€1,600-2,400
Server Chassis	Used Dell/HP/Supermicro 2U	€400-600
CPU	Xeon Silver/Gold (included)	—
RAM	128GB DDR4 ECC	€300-400
Storage	2TB NVMe SSD	€200-300
Total		€2,500-3,700

What it runs:

• All models from Budget Build (faster)
• LLaMA 2 13B with context extensions
• Mixtral 8x7B (quantized)
• Multiple concurrent inference requests

---

Efficiency Build: €3,500-5,000

Target: Always-on inference, low power consumption, quiet operation

Component	Spec	Est. Price
GPU	2x Tesla T4 16GB	€3,000-4,000
Server/Workstation	Quiet tower or 2U	€500-800
CPU	Xeon or Ryzen (included)	—
RAM	64-128GB DDR4	€200-400
Storage	1-2TB NVMe SSD	€150-250
Total		€3,850-5,450

Why this build:

• Only 70W per GPU — no special power requirements
• Can run in office environments (quiet)
• Excellent for INT8 inference workloads
• Easy to scale to 4 GPUs

---

Performance Build: €6,000-8,000

Target: Larger models, heavier workloads

Component	Spec	Est. Price
GPU	2x Tesla V100 32GB	€4,000-5,000
Server Chassis	Quality 2U/4U	€600-1,000
CPU	Dual Xeon Gold	€400-600
RAM	256GB DDR4 ECC	€600-800
Storage	4TB NVMe RAID	€400-600
Total		€6,000-8,000

What it runs:

• LLaMA 2 70B (quantized 4-bit)
• LLaMA 3 70B (quantized)
• Large context windows
• Multi-model serving

---

What’s Included

All builds include:

• ✅ Pre-installed software stack (Ollama, vLLM, or custom)
• ✅ Model downloads and configuration
• ✅ Network setup and API endpoints
• ✅ Basic monitoring dashboard
• ✅ 30-day setup support

---

← Back to localai