Gnani Artha Sovereign AI Stack Launched by Vice President C. P. Radhakrishnan

Key Facts

• Launch Date: 28 August 2026
• Launched by: Vice President of India, C. P. Radhakrishnan
• Developer: GNANI AI, Bengaluru-based startup
• Components: Evon v3.3 (Language Model) and Plexus (Enterprise AI Platform)
• Evon v3.3 Parameters: 30 billion total, ~3.5 billion active per token (Mixture-of-Experts architecture)
• Supports 11 Indian languages natively
• Open-weights model available under Apache 2.0 license via request through Hugging Face
• Designed for data residency compliance with Indian regulations like DPDP, RBI, IRDAI
• Evon v3.3 tokenizer uses ~20% fewer tokens per Indian-language word than GPT-5 tokenizers and less than half tokens compared to models like DeepSeek, Llama, and Qwen

Background & Context

India is making significant strides in developing sovereign AI technologies to retain domestic control over data and AI infrastructure. Sovereign AI stacks address high-stakes data, language, and operational complexities faced by Indian enterprises and public institutions, particularly in sectors like banking, insurance, healthcare, and telecom.

Evon v3.3 is a foundational large language model built with an efficient Mixture-of-Experts (MoE) architecture to reason and support multiple Indian languages with resource optimization. Plexus is an enterprise-grade agentic AI orchestration platform enabling deployment of autonomous AI agents designed for secure and controlled use within organizations' internal workflows.

The stack emphasizes native Indic language support (natively trained on 11 Indian languages and Indic data) to reduce the so-called 'language tax' — the additional computational cost usually incurred when processing Indian scripts. Gnani Artha's deployment within clients' own infrastructure ensures customer data never leaves the organization's network, thus satisfying regulatory requirements.

Why This Matters for Exams

The launch of Gnani Artha reflects India's efforts toward technological self-reliance, digital sovereignty, and addressing regional language diversity in AI. It is relevant to topics on India's digital initiatives, AI policy frameworks, data sovereignty laws, and indigenous technology development — important themes in current affairs and technology syllabi.

Candidates preparing for exams should understand the concept of sovereign AI stacks, the significance of indigenous large language models like Evon v3.3, and the move towards embedding AI in enterprise workflows while complying with Indian data residency regulations.

Points to Remember

• Gnani Artha was launched on 28 August 2026 by Vice President C. P. Radhakrishnan.
• Evon v3.3 is a 30-billion-parameter open-weights large language model supporting 11 Indian languages.
• Evon utilizes a Mixture-of-Experts model activating ~3.5 billion parameters per token for efficiency.
• Evon’s tokenizer is optimized for Indian scripts, reducing token usage by ~20% compared to GPT-5 tokenizers.
• Plexus is an agentic AI orchestration platform enabling autonomous, auditable AI agents across enterprise workflows.
• Gnani Artha ensures data sovereignty by enabling organizations to deploy models within their own infrastructure in compliance with Indian data residency laws.
• Model weights are available on request under Apache 2.0 license via Hugging Face.
• This initiative is part of India's broader AI self-reliance and sovereignty push under government missions like the India AI Mission.

Practice MCQs

Question 1

1. In which year was the Gnani Artha sovereign AI stack launched?
2. 2024
3. 2025
4. 2026
5. 2027

Answer: 2026

Question 2

1. Who launched the Gnani Artha AI stack?
2. Venkaiah Naidu
3. C. P. Radhakrishnan
4. M. Hamid Ansari
5. Mohammad Hamid Ansari

Answer: C. P. Radhakrishnan

Question 3

1. How many Indian languages does the Evon v3.3 model natively support?
2. 5
3. 8
4. 11
5. 15

Answer: 11