SOFTWARE MEETS MATCHED HARDWARE

The right performance for your AI.

From your first AI project to central AI infrastructure: meloki grows with your requirements. Together, we'll find the appliance that suits your models, data and workflows.

INSIDE YOUR APPLIANCE

Look under the lid.

Choose a model. Discover its components.

1U
BUILT TO WORK TOGETHER

More than the sum
of its parts.

SECUDOS software and scope7 hardware from Landitec form a matched environment for your local AI.

Start with a prepared appliance and configure models, knowledge and applications for your business. Landitec is your sales and hardware partner; the AdminPACK supports your team in day-to-day operation.

Explore the platform
01

Matched to the task

The model, context and data sources determine the resources you need.

02

For your operations

Rack format and redundant power supplies support integration into your infrastructure.

03

Planned together

We'll review the actual workloads and planned services with you.

FOUR APPLIANCES. YOUR CHOICE.

Your AI has a home.

Choose the starting point for your local AI. The exact sizing is based on your project, rather than a fixed number of users.

scope7-4480 – front view of the rack appliance
Your introduction to local AI

scope7-4480

A compact foundation for your first knowledge assistants, document analysis and local language models.

GPU
1 × NVIDIA L4
GPU memory
24 GB
CPU
AMD Ryzen 7 9700X8 cores · 16 threads
RAM
128 GB DDR5
Storage
1 × 1 TB NVMe + 1 × 1 TB 2.5″ SSD
Form factor
1U rack appliance
  • Pre-installed meloki software
  • Redundant power supplies
  • Compact rack format
Get advice on this applianceProduct details at Landitec
scope7-5580 – front view of the rack appliance
More room for your team

scope7-5580

Two GPUs and more RAM give you additional options for parallel AI services.

GPU
2 × NVIDIA L4
GPU memory
2 × 24 GB
CPU
AMD EPYC 933432 cores · 64 threads
RAM
256 GB DDR5
Storage
1 × 2 TB U.2 NVMe + 1 × 2 TB M.2 NVMe
Form factor
1U rack appliance
  • Two separate GPUs
  • Redundant power supplies
  • More resources for parallel services
Get advice on this applianceProduct details at Landitec
scope7-6680 – front view of the rack appliance
Room for demanding models

scope7-6680

96 GB of memory on one GPU for projects needing more room for models and context.

GPU
1 × NVIDIA RTX PRO 6000 Blackwell
GPU memory
96 GB
CPU
AMD EPYC 953564 cores · 128 threads
RAM
384 GB DDR5
Storage
1 × 4 TB U.2 NVMe + 1 × 4 TB M.2 NVMe
Form factor
2U rack appliance
  • 96 GB VRAM on one GPU
  • Redundant power supplies
  • 2 × 10 GbE RJ45
Get advice on this applianceProduct details at Landitec
scope7-6680-MAX – front view of the rack appliance
For particularly demanding projects

scope7-6680-MAX

Four GPUs for extensive local AI environments. The configuration is planned around your services and models.

GPU
4 × NVIDIA RTX PRO 6000 Blackwell
GPU memory
4 × 96 GB
CPU
AMD EPYC 9745128 cores · 256 threads
RAM
768 GB DDR5
Storage
2 × 4 TB U.2 NVMe + 1 × 4 TB M.2 NVMe
Form factor
2U rack appliance
  • Four separate GPUs
  • Redundant power supplies
  • 2 × 10 GbE RJ45
Get advice on this applianceProduct details at Landitec

GPU memory is shown per graphics card. Multiple GPUs do not automatically form a shared memory pool. Whether a model can be distributed across multiple GPUs depends on the model, runtime and configuration.

THE DETAILS AT A GLANCE

Compare what matters to you.

GPU memory, CPU, RAM and storage – clearly compared side by side.

The four meloki appliances compared directly
Specificationsscope7-4480scope7-5580scope7-6680scope7-6680-MAX
Graphics cards1 × NVIDIA L42 × NVIDIA L41 × NVIDIA RTX PRO 6000 Blackwell4 × NVIDIA RTX PRO 6000 Blackwell
GPU memory24 GB2 × 24 GB96 GB4 × 96 GB
ProcessorAMD Ryzen 7 9700XAMD EPYC 9334AMD EPYC 9535AMD EPYC 9745
Cores / threads8 / 1632 / 6464 / 128128 / 256
RAM128 GB DDR5256 GB DDR5384 GB DDR5768 GB DDR5
Storage1 × 1 TB NVMe + 1 × 1 TB 2.5″ SSD1 × 2 TB U.2 NVMe + 1 × 2 TB M.2 NVMe1 × 4 TB U.2 NVMe + 1 × 4 TB M.2 NVMe2 × 4 TB U.2 NVMe + 1 × 4 TB M.2 NVMe
Form factor1U rack appliance1U rack appliance2U rack appliance2U rack appliance

On small screens, you can scroll the table sideways.

YOUR AI PROFILE IN FOUR STEPS

Which appliance is right for you?

Your tasks. Your team. Your usage. Put together your AI profile and find the starting point for your consultation.

Your hardware. Chosen together.

Compare the four appliances or discuss your tasks directly with us.

The selection guide offers initial guidance. Model size, quantisation, context length, parallel requests and additional services determine actual resource needs. We'll discuss the final configuration with you.

YOUR TIME HAS VALUE.

What does
“just a quick task” cost you?

Searching, summarising, reworking: small tasks add up. Calculate what saved time could be worth to your team – compared with your AI costs.

meloki
YOUR NEXT STEP

Let's choose your hardware together.

Tell us what you want to achieve with AI. In a personal demo, we'll discuss suitable models, the resources you need and your next steps.

Arrange a demo
The meloki mascot flies towards you