Roadmap

#Instill VDP

#2023 Q1

• ✅ Add API Gateway / Reverse Proxy
• ✅ Support VDP Kubernetes deployment

#2023 Q2

• ✅ Instill Cloud - our hosted service provides you with all the power of VDP, without managing infrastructure
• ✅ No-code, drag-and-drop pipeline builder
• ✅ Flexible and complex pipelines through a DAG design

#2023 Q3 and Q4

• ✅ Introduce Numbers Protocol Connector to turn digital media files into Web3 assets
• ✅ Add a dashboard for out-of-the-box monitoring
• ✅ Logging
• ✅ Python SDK
• ✅ TypeScript SDK
• ✅ Instill CLI
• ✅ [Public pipelines for discovery and sharing
• 🔌 Support new components
  • ✅ Stability AI
  • ✅ OpenAI Text/Chat completion, Whisper, and Embeddings
  • ✅ Hugging Face
  • Assembly AI
  • ✅ Pinecone
  • ✅ BigQuery
  • ✅ Google Cloud Storage (GCS)
  • AWS S3
  • Snowflake
  • Milvus
  • ✅ Base64 operator to encode and decode Base64 data
  • ✅ JSON operator to manipulate JSON data to/from string or with jq command
  • ✅ Image operator for image visualization
  • ✅ Text operator to convert document into text, split text into chunks, and more

#2024 Q1 and Q2

• 🔌 Support new components
  • Gemini
  • Iterator component to manipulate an array of data elements
  • Component to filter an array of data elements
  • VDP-in-VDP component for encapsulation
• 🎯 Failover
• 🎯 Component versioning
• 🎯 Scheduled pipeline

#Instill Model

Instill Model is under rapid development. Our team is pushing new features every day to meet the needs of our users. Here is our public high-level roadmap.

#2023 Q2 and Q3

• ✅ Instill Cloud: our hosted service provides you with all the power of Instill Model, without managing infrastructure
• ✅ Decouple model deployment feature from VDP and introduce a standalone ModelOps platform called Instill Model, providing a more stable model inference experience

#2023 Q4

• ✅ Serve Pre-trained models
  • Llama 2
  • LLaVA
  • Zephyr
  • Stable Diffusion XL (SDXL)
  • ControlNet

#2024 Q1 and Q2

• 🎯 Serverless model serving
• 🔮 Custom model training: Llama-2
• 🔮 Custom model training: Stable Diffusion

#Contributing to the Roadmap

Your feedback is important for us to understand how to build better and more valuable tools for the community. If you'd like to send us a feature request, have a look at the issues first. If your idea has already made it to the planning, give it the thumbs up 👍 to vote for it. If not, you can create a new issue. Besides, you can also join our community and request it in the #instill-model channel.