Serverless Models
Gravix Layer provides instant access to high-performance AI models through our serverless infrastructure. Deploy powerful language and vision models without managing any infrastructure.
Text Models
| Model Name | Model ID | Organization | Parameters | Context Length | Quantization |
|---|---|---|---|---|---|
| Llama-3.1-8B-Instruct | llama3.1:8b-instruct-fp16 | Meta-Llama | 8B | 128,000 | FP16 |
Features:
- Optimized for instruction following and conversational AI
- Extended context window for processing large documents
- Fine-tuned for tool usage and function calling
- High-performance FP16 quantization for optimal speed
Vision Models
| Model Name | Model ID | Organization | Parameters | Context Length | Quantization |
|---|---|---|---|---|---|
| Qwen2.5-VL-3B-Instruct | qwen2.5vl:3b-fp16 | Qwen | 3B | 32,768 | FP16 |
Features:
- Multimodal processing combining text and images
- Efficient 3B parameter architecture for fast inference
- Advanced visual reasoning capabilities
- Support for image analysis and description tasks
Embedding Models
Status: Coming Soon
Advanced embedding models for semantic search and vector operations are currently in development and will be available soon.
Model Selection Guide
| Use Case | Recommended Model | Best For |
|---|---|---|
| Text Generation | Llama-3.1-8B-Instruct | Conversational AI, content creation, code generation |
| Document Analysis | Llama-3.1-8B-Instruct | Large document processing with 128k context |
| Image Analysis | Qwen2.5-VL-3B-Instruct | Visual question answering, image description |
| Multimodal Tasks | Qwen2.5-VL-3B-Instruct | Combined text and image processing |