A Sovereign, Indic-First, Reasoning Model
Gnani-Evon-v3.3-30B-A3B is a long-context hybrid Mamba-Transformer mixture-of-experts model for reasoning, instruction following, and tool use in English and eight Indian languages.
ProofPaper ↗
Key points
- Starting from NVIDIA's Nemotron-3-Nano-30B-A3B checkpoint, it retains the 52-layer backbone and changes the vocabulary, training data, and post-training objectives.
- Continual pretraining then adds broad English-Indic data, high-quality annealing, and direct training at a 131,072-token sequence length.
- Reinforcement learning across four verifiable task environments improves mathematics, graduate-level science, and instruction following, but degrades Indic extractive question answering because the rewards do not require an answer in the requested language.
- Relative to the architecture-identical starting checkpoint, the released model gains 5.2 points on an Indic aggregate while giving up 1.7 points of English retention.
Sources (1)
- [1]A Sovereign, Indic-First, Reasoning ModelarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 11, 05:52 AM
Gnani-Evon-v3.3-30B-A3B is a long-context hybrid Mamba-Transformer mixture-of-experts model for reasoning, instruction following, and tool use in English and eight Indian languages.
Starting from NVIDIA's Nemotron-3-Nano-30B-A3B checkpoint, it retains the 52-layer backbone and changes the vocabulary, training data, and post-training objectives.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 11, 2026Apertus 1.5 Release Notes
- Oct 10, 2026Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
- Oct 10, 2026VC-OPD: Visual Counterfactual On-Policy Distillation for Grounded Vision-Language Reasoning
- Oct 9, 2026Qwen/Qwen-Image-2.1-Turbo
- Oct 9, 2026Introducing Clef-omni with full multimodality, plus a faster Clef and a cheaper Clef-flash
- Oct 9, 2026Rethinking Relative Policy Differences Transfer of On-Policy Distillation