SuperMap iPortal AI Assistant supports Windows and Linux operating systems. For graphics cards, it supports NVIDIA, AMD, and Moore Thread GPUs. NVIDIA GPUs are recommended as AMD GPUs perform slower than NVIDIA equivalents.

Minimum Requirements (Qwen/Qwen3-8B)

• RAM: 32GB
• GPU Model: NVIDIA RTX 4090
• GPU Memory: Single or multiple GPUs with total memory ≥16GB (e.g., NVIDIA RTX 4070 Ti or equivalent)
• Processor: Dual-core 2.30 GHz CPU
• Storage: SSD with ≥40GB free space recommended

Recommended Configuration (Qwen/Qwen3-14B)

• RAM: 32GB or higher
• GPU Memory: Single or multiple GPUs with total memory ≥48GB (e.g., A100, H100, dual NVIDIA RTX 4090, or equivalent)
• Processor: Intel Core i7 or higher
• Storage: SSD with ≥100GB free space recommended

Performance Metrics (Qwen/Qwen3-14B)

Performance test results for the Qwen/Qwen3-14B model:

• Short queries: about 40 tokens/second response generation
• Medium conversations (about 6,000 tokens): about 100 tokens/second response generation
• Long conversations (about 14,000 tokens): about 130 tokens/second response generation
• Complex scenarios (about 30,000 tokens): about 150 tokens/second response generation
• Default instruction capacity: Pre-configured with full instruction/tool sets (about 20,000 tokens), eliminating context length concerns
• Tool invocation efficiency: Maintains 130-150 tokens/second response speed with all tools enabled

Note:

Generally, 1 token ≈ 3-4 English characters or 1.5-1.8 Chinese characters.