India Launches Gnani Artha, a Government‑Backed Sovereign AI Stack

Overview

• GNANI AI and Vice‑President CP Radhakrishnan unveiled Gnani Artha at Uprashtrapati Bhavan on Friday, August 28, 2026, presenting it as a government‑backed step toward domestic AI capability.
• The stack combines Evon v3.3, a 30‑billion‑parameter open‑weights language model trained natively in 11 Indian languages, with Plexus, an agentic platform meant to link AI to institutional workflows.
• GNANI AI says Evon v3.3 weights are available under an Apache 2.0 licence and that the model was trained for reasoning and Indic performance, with plans to expand to larger model sizes and more languages.
• Company leaders claim Evon uses about 40% fewer tokens on Indian‑language workloads and has shown early interest from banks, insurers and retailers, though independent verification of those efficiency and adoption claims was not provided.
• Officials framed the stack as part of the India AI Mission push for technological self‑reliance and pointed to existing public‑sector uses such as Digital Sansad translations and recent eight‑language AI interpretation in Parliament, with GNANI noting future vertical engines and speech‑to‑speech work as next steps.