Skyfall AI introduces SKY-001, an engineering world model for CAD
A team of former Maluuba specialists has launched Skyfall AI with SKY-001, a model that turns text prompts or sketches into complete CAD projects.
Microsoft acquired Maluuba a year before GPT-1 was released. The company's founder admits he regrets the sale, because Maluuba could have become the next OpenAI.
The same deep-learning team has returned with a new project, Skyfall AI, and its first model is not just another LLM.
SKY-001 is the first engineering world model. You can give it a text prompt or a rough sketch, and it will design the entire object, break it down into individual parts down to screws, and then recreate it as a full CAD file that can be opened in programs such as Autodesk Fusion.
According to the company's tests, for CAD and BIM tasks the model works 100 times faster and is 150 times cheaper than GPT-6 Astra.
Why it matters
AnalysisSKY-001 could significantly speed up engineering design and reduce CAD/BIM costs by offering a specialized model instead of a general-purpose LLM.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Odyssey shows a fly in its AI simulator Odyssey-3
Startup Odyssey added a fly model to its AI simulator Odyssey-3, where the insect moves autonomously through a generated world while video is produced online in real time.
Jev and System One AI: A Complete Practical Guide to the Decision Model from TypeSafe
A deep engineering breakdown of the flagship Jev model: System One architecture versus LLMs, Choice/Score/Noul primitives, request parallelism, confidence mathematics, video use cases, code, and an overview of local open-weight alternatives (Laya, GLiNER2.5, CLM-8B).
Part II. AI Agent Architecture: From Model to Memory and Tools
A deep engineering dive: tokenization, KV-cache, sampling determinism, the training stack, quantization, context engineering, the safety stack, production RAG, ReAct, and multi-agent swarm systems.
Jev Decision Model
Jev is a commercial non-autoregressive System 1 model from TypeSafe AI for fast typed decisions in agentic loops: it returns a validated decision object in 70–500 ms at ~$0.042/1M input tokens, eliminating JSON syntax hallucinations and high-latency routing in ReAct loop/Tool Calling stacks.
MCP (Model Context Protocol)
An open standard from Anthropic based on JSON-RPC 2.0 for unified bidirectional connection of AI assistants to external tools, databases, and system environments.
Anthropic Reportedly Preparing Fable 5.5 Model
Insiders who tested a new Anthropic model say it better understands user intent from short prompts and is especially strong at finding bugs and vulnerabilities.