On-Premises Enterprise-Grade Server
Deepseek AI Model Solution
For running artificial intelligence and generative AI
Technical Specifications
CPU
2 x AMD EPYC 7513 (2.6GHz /32C)
Memory
16 x 32Gb DDR4 RECC
Storage
2 x 480Gb SATA SSD
Operation System
Ubuntu 22.04.5 LTS or above
Display Card
4 x Geforce RTX4090 24Gb Turbo
Power Supply
1600W
Port
25Gb Ethernet Dual-Port SFP+
Running Model
DeepSeek-R1Distill-Llama-70B-FP8-Dynamic
Inference Engine
vLLM
Safe
Fast
Better
Secure Computing
Let data within control
Keep AI models and critical data within your organization—no need to hand them over to third-party AI processing and take on unnecessary risks. Every data boundary can be defined, validated, and audited. Enhanced security isn't just a slogan; it's a practical, implementable solution.
Data Sovereignty: All raw data, temporary storage, and usage logs remain on local servers, preventing confidential data leaks.
Encrypted Transmission: All internal and external traffic is encrypted via TLS, safeguarding organizational assets.
Efficiency
Leverage A.I. models to boost productivity
Combining AMD EPYC processors, NVIDIA GPUs, and next-generation memory, storage, and expansion technologies to fully unleash computational potential. This not only significantly reduces time-to-market but also delivers uncompromising AI acceleration performance.
Distilled from the DeepSeek-R1 architecture and the larger Llama 3.3 70B-Instruct inference model, it excels in powerful reasoning, mathematical operations, and program generation, while also mastering output tasks like summarization and translation.
Rapid Deployment
on your own AI platform
Applicable to multiple scenarios: Enables fast retrieval of internal corporate documents, assists with code generation, document translation, and more.
Flexible application: Supports multiple languages and scenarios, meeting cross-departmental and cross-business needs.
Supports multilingual content generation and localized translation, enabling multilingual market promotion.
Functions
Generate Creative Ideas
Assist in writing articles, marketing copy and replying to emails.
Translation Capabilities
Multi-language support, including Chinese and English.
Rapid document Analysis
Analyse complex documents and summarise key points to improve productivity.
Smooth Operations
No steep learning curve with dialogue-based operations.
Various Input Methods
Audio, pictures, and computer files can be easily used as an input method.
Instant Response
Our selected models could provide quick answers to complex questions.
