KiteFishAI launches NanoI1 model for local enterprise AI tasks

What happened
KiteFishAI, a startup founded by former Accenture AI architect Anuj Gupta, has released NanoI1, a 155-million-parameter language model. The model is designed for enterprises to perform AI tasks such as document processing and customer support locally on their own infrastructure. NanoI1 supports multiple Indian languages and can handle a 2,048-token context window. The company developed the model with support from government GPU grants and aims to provide cost-effective AI solutions for organizations with limited computing resources.
Key details
- NanoI1 has 155 million parameters and a file size of 311 MB.
- NanoI1 supports English, Hindi, and eight other Indian languages.
- It is available under the Apache-2.0 license in both Base and Instruct versions.
- KiteFishAI received ₹50 lakh in government support and spent about ₹4 lakh on NanoI1's training and development.
Why it matters
Indian enterprises can now deploy AI for document extraction, classification, and customer support without relying on large, expensive cloud-based models. Organizations with limited IT budgets or strict data privacy requirements can run NanoI1 on their own servers or private cloud, reducing costs and keeping sensitive data in-house. This gives companies a new, affordable option for automating routine tasks while maintaining control over their infrastructure and data.

