Sarvam’s Token Factory is now live. It offers Indian enterprises access to eight distinct open-source large language models through GPU clusters hosted entirely within India. This launch, announced today by Inc42 and MarkTechPost, establishes new infrastructure for India’s artificial intelligence sector. It provides a local alternative for AI inference, with implications far beyond commercial convenience.
The Token Factory makes models like Sarvam 105B, GLM 5.3, DeepSeek V4.1 Flash, Kimi K3.1 Pro, and Gemma 4 31B available through a single, OpenAI-compatible API. This approach simplifies integration for developers already familiar with popular AI interfaces, while addressing a fundamental need for data sovereignty and reduced latency in AI deployments across India.
From Defence Contracts to Public Infrastructure
This move follows a clear pattern of strategic development for Sarvam. Just weeks ago, on October 27, 2026, the company secured a high-profile partnership with the Ministry of Defence to drive AI adoption for the armed forces, as Inc42 and Analytics India Magazine reported. That five-year deal covers AI implementation across the Army, Navy, and Air Force. For an entity like the Ministry of Defence, data processing and model inference location is a national security imperative. The Token Factory explicitly hosts its inference capabilities on GPU clusters within India, directly addressing this need for sovereign AI infrastructure.
India-hosted inference holds strategic importance for sensitive defence applications. Processing classified information or strategic intelligence through models running on servers outside national borders presents significant risks. Sarvam’s decision to build out domestic GPU infrastructure and offer a local inference service aligns with broader national security objectives. It provides a trusted environment for critical AI workloads.
Broadening Access and Competing on Infrastructure
Beyond government contracts, Sarvam has also been positioning itself in the broader enterprise AI market. On November 5, 2026, Inc42 noted Sarvam’s “Prism” offering entering India’s “AI decision model race.” While the specifics of Prism remain less detailed than the Token Factory, its mention indicates Sarvam’s intent to develop and deploy its own AI capabilities for business decision-making. The Token Factory appears to be a complementary, perhaps even foundational, platform that democratizes access to a suite of models, both Sarvam’s own (like Sarvam 105B) and other prominent open-source options.
The availability of eight open models, accessible via a standard API, reduces the barrier to entry for Indian startups and established enterprises looking to integrate advanced AI into their operations. Instead of managing complex GPU infrastructure, licensing individual models, or dealing with the overhead of multiple vendor APIs, businesses can tap into a consolidated service. This positions Sarvam as a key player in India’s cloud AI infrastructure, competing on model quality, efficiency, and localized control of its compute resources. The API’s OpenAI compatibility ensures developers can switch or integrate with minimal friction. This uses a widely adopted standard.
Sarvam’s strategy appears to be one of building foundational components for India’s AI development. High-assurance government contracts provide a strong initial use case for sovereign AI infrastructure. Expanding this to a public-facing inference platform, hosted locally, extends these benefits to a wider array of businesses and developers. This trajectory suggests the company aims to enable AI adoption in India and provide specific AI applications. The investment in GPU clusters, while not quantified in the available research, shows a substantial commitment to physical infrastructure. This component of AI at scale is critical but often overlooked.
The Path Ahead for India’s AI Ambitions
The Token Factory launch shows India’s AI sector maturing. It moves beyond mere application development, focusing on foundational infrastructure. For India’s technology ambitions, particularly its push for self-reliance in critical sectors, home-grown inference capabilities are essential. This reduces reliance on global cloud giants for sensitive data processing and supports local AI development and deployment. Sarvam’s trajectory, from securing a major defence partnership to launching a broad inference platform, suggests a deliberate effort to establish itself at the heart of India’s strategic AI capabilities.
Quick Facts
What is Sarvam Token Factory?
Sarvam Token Factory is a platform that provides India-hosted inference for eight open-source large language models through GPU clusters in India, accessible via an OpenAI-compatible API.
Which models does Sarvam Token Factory support?
Sarvam Token Factory supports Sarvam 105B, GLM 5.3, DeepSeek V4.1 Flash, Kimi K3.1 Pro, and Gemma 4 31B, among others.
What was Sarvam’s previous significant announcement?
On October 27, 2026, Sarvam partnered with the Ministry of Defence for a five-year deal to drive AI adoption for the Indian armed forces.
Compiled by Launch91 Desk from the sources linked above. More about Launch91.