The company announced that over 35 million pages are presently being digitised via the Sarvam Vision API, prompting the decision to share these efficiency benefits with customers through a notable price reduction.
Launched in February, Sarvam Vision is a vision-language model tailored for document intelligence and digitisation processes. Since its release, the platform has managed millions of documents, allowing organisations to convert physical records into searchable and easily accessible digital formats.
In a recent post on X, the company disclosed that it is lowering the cost of the Sarvam Vision API from ₹1.5 per page to ₹0.5 per page, owing to enhanced efficiency achieved at scale since its launch.
The post states, “Earlier this February, we launched Sarvam Vision, a vision-language model for document intelligence. Today, more than 35 million pages are being digitised via the Sarvam Vision API by developers and partners. Since launch, we’ve significantly improved our efficiency to operate at scale, and we’re now passing these advantages on by reducing the Sarvam Vision API price from 1.5 to 0.5 per page.”
Earlier this February, we launched Sarvam Vision, a vision-language model for document intelligence.
Today, more than 35 million pages are being digitised through the Sarvam Vision API by developers and partners.
Since launch, we’ve made it significantly more efficient to serve… pic.twitter.com/iqjEbZeNGF
— Sarvam (@SarvamAI) May 29, 2026
The adjusted pricing represents nearly a 67% reduction, making the platform considerably more cost-efficient for enterprises and developers engaged in large-scale document processing.
Clarifying the changes, the company elaborated, “As adoption surged, we began catering to users with much higher document volumes. Thus, we revamped sections of our serving stack – optimised inference kernels for state-space architecture, smarter page-level batching, and improved hardware utilisation across our sovereign cloud. The end result is a model that operates far more efficiently at scale. We’re passing these benefits directly to our users through this price reduction.”
Sarvam Vision is engineered for document intelligence and OCR (optical character recognition) tasks, enabling users to extract and digitise information from scanned documents, forms, archives, handwritten manuscripts, and financial records. Crafted for India’s multilingual landscape, the model accommodates all 22 official Indian languages.
The company previously indicated that Sarvam Vision achieved an accuracy score of 84.3% on olmOCR-Bench and 93.28% on OmniDocBench v1.5 during benchmark assessments.
Positioning itself as a sovereign AI entity, Sarvam AI is dedicated to creating foundational AI technologies for India while utilizing domestic infrastructure. Its overarching aim is to make advanced AI capabilities more accessible to Indian developers, enterprises, and institutions.
This latest price revision coincides with a burgeoning demand for AI-driven document processing solutions across various sectors, including finance, healthcare, education, and public services.
