RESEARCH
Wepushbeyondthestateoftheart.
SAGEA is a pioneer frontier AI research lab from Nepal, advancing research across language, reasoning, multimodality, and intelligent systems.
Our models
Explore our model collectionSAGE 2.6 Celer
Hybrid MoEA compact hybrid instruct-and-reasoning MoE with multimodal input. Unifies instruction, reasoning, and agentic capabilities in a single small model, achieving state-of-the-art performance among open weights on coding, finance, and manufacturing domains.
Publications.
Discover how our models and methods advance reasoning, perception and action.
Filter by topic 6 results
SAGE 2.6 Celer
Advanced language model with enhanced reasoning capabilities.
FOUNDATION MODELS
REASONING
GLUE Engine
Advancing OCR for Devnagari and other scripts with improved accuracy.
FOUNDATION MODELS
VISION
SAFETY
SAGE Actus 2.4
Next-generation model architecture for improved performance.
APPLIED AI
FOUNDATION MODELS
SAGE Nano: Thinking About Thinking
Exploring meta-cognitive capabilities in AI systems.
REASONING
CODING
Fragile Mastery
Understanding the limits and robustness of AI model capabilities.
SAFETY
FOUNDATION MODELS
Rosetta Paradox
Investigating cross-lingual transfer and language understanding.
VISION
REASONING
Research areas.
Reasoning and multimodality
Reason across language, media, and code.
Document intelligence.
Models that excel at parsing complex documents.
Identity verification.
Models that confirm who someone is from faces, voices, and ID documents.
Safety and guardrailing.
Models that evaluate text and images against policies written in natural language.
Simulation.
Complex environment simulation, from automotive airflow and plasma turbulence to industrial particle flows.
Audio.
Multilingual speech understanding, generation, and adaptation.
Inside the work.
Go deeper on the techniques, systems, and engineering behind our models, from training and evaluation to inference and efficiency.
SAGE 2.6 Celer
Advanced language model with enhanced reasoning capabilities.
ARVA OCR: High-Fidelity Document Parsing
State of the art document intelligence model for Nepali and Devanagari scripts.
GLUE Engine: OCR for Devanagari Scripts
Solving the Ransom Note effect in Devanagari synthetic data generation.
SAGE 2.4 Actus
Next-generation model architecture for improved performance.
Mira Reasoning
Deliberate, step-by-step reasoning model for complex problem-solving.
ARVA 2 OCR
Advanced document intelligence with OCR capabilities.
Vaani TTS
Natural text-to-speech synthesis system.
Build on our open-weight models.
Download, adapt, and deploy our models.
Discover SAGEA on Hugging Face
Discover SAGEA on Github
We’re seeking researchers and engineers to advance the models, systems, and techniques behind SAGEA.
