Models

All models are open source on Hugging Face. Trained on our GPU cluster.

NanoVLA-Flow-V1

Parameter-Efficient Continuous Vision-Language-Action Model

Vision-language-action model for robotic control built on Gemma-4-E2B with a Flow Matching action head. Predicts continuous 3D velocity trajectories instead of discrete action tokens. Trained on a single 16GB T4 with LoRA, weights merged for inference.

AgentGuard-2.8B

Local AI Agent Security via Mamba-2

Mamba-2 SSM fine-tuned to detect prompt injection, exfiltration, and tool-call hijacking in AI agent sessions. Runs as a local sidecar with O(1) memory — monitors arbitrarily long agent trajectories without truncation.

CBD-LLM-PoC-V1

Causal Block Diffusion for Large Language Models - Proof of Concept v1

Hybrid diffusion architecture enabling block-parallel text generation while retaining standard causal attention and KV caching.

Qwen3-0.6B-Tool-Router

Efficient Tool Routing System

Lightweight tool call router for agentic systems. 29.2% accuracy overall vs industry 23.93 600M params models on BFCL

IND-QWENTTS-V1

Indian Language Text-to-Speech v1

Multilingual TTS for 2 languages. MOS 3.8/5.0. Cross-lingual transfer from high-resource anchors. Edge-deployable.

STRM-4B-v1

Stateful Reasoning Model

LoRA fine-tune of Qwen3-4B for parsing unstructured spoken-language input into structured JSON. Maintains running state to handle corrections, cancellations, and quantity changes in a single forward pass. ~94% exact-match accuracy.