Dstack
Open-source cloud-agnostic platform for AI/ML workload orchestration
Dstack
Open-source cloud-agnostic platform for AI/ML workload orchestration
Dstack is an open-source platform for running AI and ML workloads — training runs, fine-tuning jobs, and inference services — across any cloud provider or on-premises GPU infrastructure through a unified interface. Teams define their environment requirements (GPU type, memory, Docker image) in a simple YAML configuration, and Dstack provisions the appropriate infrastructure, manages job queuing, and handles interruption recovery for spot instances automatically. This allows ML teams to use the cheapest available GPU capacity without managing cloud-specific APIs for each provider.
Key Features
- ✓Cloud-agnostic
- ✓Spot instance management
- ✓GPU orchestration
- ✓Open source
- ✓YAML configuration
- ✓Multi-cloud
Quick Info
- Category
- AI Infrastructure & MLOps
- Pricing
- Free
More AI Infrastructure & MLOps Tools
Tigris Data
AI Infrastructure & MLOpsAI-native object storage with built-in vector search and S3 compatibility
Superlinked
AI Infrastructure & MLOpsVector compute framework that helps ML engineers build retrieval systems by combining multiple data types a…
Qdrant Cloud
AI Infrastructure & MLOpsManaged vector database cloud service offering high-performance similarity search with filtering, payload i…
Weaviate Embedded
AI Infrastructure & MLOpsEmbedded mode of the Weaviate vector database that can be instantiated directly within a Python application…