New Delhi, Sept. 9 -- AI and Cloud startup Neysa.ai has expanded its open-weights catalogue to include GPT-OSS models. Neysa claims that the new addition will help Indian organisations to have full control over their AI and reduce black-box dependency on foreign-hosted platforms.
To be sure, open weights refer to the final weights and biases of a trained neural network. These values help determine how the model interprets input data and generates outputs. GPT-OSS is one such open-weight language model offering from OpenAI.
GPT-OSS will be added to Neysa's AI-native cloud platform called Velocis, alongside Llama, DeepSeek, Qwen, and Mistral. Models from all of these open-weights publishers are now available on Neysa's Velocis Cloud as Managed Inference Endpoints.
"India cannot afford to build its AI future on platforms it does not control. The size of our digital economy, the sensitivity of our data, and the urgency of innovation demand indigenous infrastructure. By bringing world-class models like GPT-OSS into our Velocis ecosystem, we are proving that advanced AI can be hosted and scaled entirely within India," said Sharad Sanghi, Co-founder and CEO of Neysa.
He also added that this announcement will give Indian enterprises the freedom to innovate while protecting their IP. It will also give better compliance with Indian regulations.
Developers will now get access to models like GPT-OSS-120B, a Mixture-of-Experts architecture for state-of-the-art reasoning, and GPT-OSS-20B, optimsed for high-performance deployments. It also includes specialised, fine-tuned versions such as GPTQ, AutoRound, and 4-bit models for custom workloads. The Velocis platform enables rapid deployment from concept to production API, with managed GPU orchestration and a unified API to switch between models.
Published by HT Digital Content Services with permission from TechCircle.