Indian Entrepreneur’s SKY-001 AI Model Claims 150x Lower Cost Than GPT-6 Astra
Indian Entrepreneur’s SKY-001 AI Model Targets Engineering, Claims 150x Lower Cost Than GPT-6 Astra
An Indian entrepreneur has introduced a new artificial intelligence model designed specifically for engineering and design tasks. Called SKY-001, the model has been developed by Skyfall AI, a company co-founded by Sam Pasupalak.
According to Pasupalak, SKY-001 can deliver competitive results on selected computer-aided design (CAD) benchmarks while operating at a substantially lower cost than some leading AI models, including GPT-6 Astra.
The model is the result of approximately 18 months of work by the Skyfall AI team. Its approach focuses on breaking complex engineering projects into smaller, structured steps rather than trying to generate an entire design in one go.
What Is SKY-001 AI?
SKY-001 is an AI model developed for engineering design tasks, including computer-aided design (CAD) and building information modelling (BIM).
CAD software is used to create and modify technical drawings, components and three-dimensional models. BIM, meanwhile, helps professionals develop digital representations of buildings and organise information about their structures and individual components.
These tasks can be challenging for AI systems because a design may involve numerous connected parts, precise dimensions and dependencies. A mistake in one component can affect the overall result.
Skyfall AI says SKY-001 is designed to address this challenge through a more structured process that divides a complex project into smaller tasks.
How Does SKY-001 Work?
Instead of generating a complete engineering design all at once, SKY-001 reportedly plans the project in stages. It first breaks a design into assemblies and individual components before working through the steps needed to construct them.
The process can involve visual representations, sketches and a sequence of construction steps. By organising the work into smaller units, the model aims to maintain consistency between individual components and the overall design.
A similar approach can be used for building design. In BIM-related tasks, the model can start with a building outline and divide it into individual units and rooms.
According to Skyfall AI, this method is intended to support complex projects involving hundreds or even thousands of steps. Its real-world effectiveness will depend on the accuracy of the output and how well the model performs across different engineering tasks.
SKY-001 vs GPT-6 Astra: What Do the Reported Benchmarks Show?
Pasupalak has shared benchmark results comparing SKY-001 with other AI models. He claims that SKY-001 is between 51 and 125 times faster than the frontier models tested, while costing between 150 and 210 times less, depending on the comparison.
One of the reported CAD cost comparisons lists the following figures:
| Metric | SKY-001 | GPT-6 Astra |
|---|---|---|
| Reported cost per sample | $0.005 | $0.75 |
| Reported IoU score | 79.7% | 79.1% |
Based on these figures, the reported cost of the GPT-6 Astra run is 150 times that of SKY-001 for the particular comparison. The accuracy difference is much smaller, with SKY-001 recording 79.7 per cent against 79.1 per cent.
What Is IoU and Why Does It Matter?
IoU stands for Intersection over Union. It is a metric that measures the overlap between a predicted result and a reference result.
In design-related tasks, it can help evaluate how closely an AI-generated output matches the expected geometry or reference representation. A higher score generally indicates greater overlap under the benchmark's evaluation method.
However, IoU is only one measure of performance. Engineering designs may also need to meet requirements related to dimensions, structural constraints, component compatibility and manufacturability. A strong score on one benchmark does not automatically prove that a model is better for every engineering application.
Are SKY-001’s Performance Claims Independently Verified?
The speed, cost and accuracy figures discussed here come from Skyfall AI's reported tests and the results shared by its founder. They have not been independently verified in the information supplied for this article.
The comparisons relate to particular tasks and testing conditions. They do not establish that SKY-001 outperforms general-purpose AI models across all types of work.
Actual costs and performance can vary depending on the task, hardware, model configuration and evaluation method. Independent testing across a wider range of engineering workflows would help provide a clearer picture of the model's capabilities.
Who Is Sam Pasupalak?
Sam Pasupalak is an entrepreneur from Bhubaneswar, Odisha, who is based in San Francisco, United States. He is the co-founder and CEO of Skyfall AI, the company developing SKY-001.
According to the supplied report, Pasupalak completed a bachelor's degree in mathematics and computer science at the University of Waterloo in Canada in 2011. He later earned a master's degree in political science and government from the London School of Economics in 2019.
His academic experience at the University of Waterloo included areas such as artificial intelligence, robotics and human-computer interaction.
From Maluuba to Skyfall AI
Before launching Skyfall AI in 2025, Pasupalak co-founded Maluuba in 2011 and served as its CEO. The AI research startup was acquired by Microsoft in 2017.
At Maluuba, he led researchers and engineers working on artificial intelligence and language understanding. His experience in AI research and entrepreneurship forms part of the background to his latest work on engineering-focused AI systems.
Why Engineering-Specific AI Models Matter
General-purpose AI systems can help with writing, coding, research and design-related tasks. However, specialised engineering work often requires structured outputs, consistency between components and close attention to technical constraints.
An AI model designed around a particular workflow may be able to focus its capabilities on those requirements. If SKY-001 can deliver reliable engineering results at the reported cost, it could be relevant to teams looking to automate parts of their design process.
Still, professional engineering applications require careful validation. AI-generated designs should be reviewed against applicable technical standards and tested before being used in real-world projects.
Conclusion
SKY-001 is an engineering-focused AI model developed by Skyfall AI, co-founded by Indian entrepreneur Sam Pasupalak. The model uses a staged approach to CAD and BIM tasks, breaking complex projects into smaller components and steps.
Its founder claims that the model can run substantially faster and at a much lower cost than the AI systems tested. The reported CAD comparison shows a small difference in IoU scores, alongside a large difference in cost per sample.
However, these figures are based on company-reported benchmarks and should not be treated as proof of broad superiority. Independent evaluations will be important in determining how well SKY-001 performs across real engineering workflows.
Disclaimer: Performance, speed and pricing figures in this article are based on the supplied report and claims attributed to Skyfall AI's founder. They have not been independently verified here and may vary across tasks and testing conditions.
Comments
Post a Comment