Blueprints
Blueprints are ready-to-use templates that help you kickstart your AI projects quickly. They include pre-configured settings for specific use cases, reducing setup time and ensuring best practices.
Overview
Blueprints provide:
- Pre-configured Templates: Optimized for specific AI workloads
- Quick Start: Deploy in minutes with one click
- Test Before Deploy: Try models directly in the browser
- Categories: Browse by Text, OCR, or Parabricks
Accessing Blueprints
Navigate to Serverless GPU > Blueprint to browse available templates.
Search and Filter
- Search: Find blueprints by name
- Categories: Filter by All, Text, OCR, or Parabricks
Available Blueprints
Text Models
| Blueprint | Description | Tags |
|---|---|---|
| Qwen3 30B A3B | Advanced text generation model with 30B parameters, optimized for natural language understanding and generation tasks | LLM, Text Generation, NLU |
| Gemma3-27b | High-performance text model for various NLP tasks including summarization, translation, and question answering | LLM, NLP, Multilingual |
| Typhoon2.1-gemma3-12b | Thai-English bilingual model based on Gemma architecture, specialized for Southeast Asian languages | LLM, Thai, Bilingual |
OCR Models
| Blueprint | Description | Tags |
|---|---|---|
| Thai Document OCR | Optical character recognition model supporting Thai and English text extraction from images and documents | OCR, Vision, Document AI |
Genomics
| Blueprint | Description | Tags |
|---|---|---|
| Parabricks fq2bam | GPU-accelerated FASTQ to BAM alignment using NVIDIA Parabricks fq2bam | Genomics, Alignment, GPU |
Using Blueprints
Browse Blueprints
- Navigate to Serverless GPU > Blueprint
- Browse available templates or search by name
- Filter by category (Text, OCR, Parabricks)
- Click on a blueprint to view details
View Blueprint Details
Each blueprint page includes:
- Description: What the model does and its capabilities
- Tags: Categories and use cases
- Chat Interface: Test the model before deploying
- Documentation: Usage instructions and examples
Deploy a Blueprint
- Click on a blueprint to open the details page
- Test the model using the chat interface (optional)
- Click Deploy Blueprint
- The blueprint will be deployed to your Serverless GPU project
Test Before Deploying
Each blueprint includes a chat interface where you can:
- Send messages to test the model
- Evaluate response quality
- Verify the model fits your use case
Blueprint Categories
Text
Text blueprints are optimized for:
- Natural language understanding
- Text generation
- Summarization and translation
- Question answering
- Thai and multilingual support
OCR
OCR blueprints provide:
- Thai and English text extraction
- Document processing
- Image to text conversion
Parabricks
Parabricks blueprints enable:
- GPU-accelerated genomics pipelines
- FASTQ to BAM alignment
- DNA/RNA sequencing workflows
Pricing
Blueprints use Serverless GPU pricing:
| Instance | On-Demand | Spot (Save 50%) | Storage |
|---|---|---|---|
| H100 | $4.32/hr ($0.0012/sec) | $2.16/hr ($0.0006/sec) | $1.00/GB/mo |
View current pricing at Serverless GPU > Pricing.
Next Steps
- Serverless GPU - Learn more about Serverless GPU
- One-Click Deployment - Deploy vLLM models on GPU Instance