Cantonese Large Language Model

Our self-developed Cantonese LLM powers Speech-to-Text (STT), Text-to-Speech (TTS), Speech-to-Speech, and Text-to-Text. Models specifically designed to understand, interpret, and generate authentic Cantonese, bringing 86 million speakers into the GenAI era.

Cantonese LLM
Community Version

Open-Sourced

Built with hon9kon9ize, a Hong Kong AI research community, our Cantonese LLM embraces an open-source, open-data, open-model, and open-community approach.

LLMs for
Low-Resource Languages

At Votee.ai, we're building the next generation of AI solutions – models that truly understand context. That's why our mission is clear: to push the boundaries of low-resource LLMs capturing the intricate ways locals actually use their language. We believe authentic communication, powered by true contextual understanding, fuels better business.

Request a Demo
World map illustrating Votee's focus on developing Large Language Models (LLMs) specifically for low-resource languages across the globe.
Source: https://www.nature.com/articles/s41559-021-01604-y
High-Quality Data

Drawing data from diverse, proprietary channels like Votee's advanced social listening and survey platform,  academic research, and extensive public Cantonese datasets.

Supports Multiple Architectures

We have successfully completed large model training on architecture like LLaMa-3, Yi, Qwen, and Gemma 2. More on the horizon.

Versatile Applications

Fluid intelligent communication for any environment - from the formality of a boardroom to the dynamic conversations in classrooms, and even late-night snack orders.

Scalability

Designed with HK-level hustle. We handle immense volumes, processing more chats than the MTR has rush-hour patrons.

Revolutionizing WhatsApp Chats with Cantonese LLM: SANUKER X VOTEE

As leading AI solutions providers in Hong Kong, Votee and Sanuker bring our Cantonese LLM to WhatsApp for easy enterprise AI communication.

Version Comparison

Not All 廣東話 Models Are Created Equal
Features
Community
Enterprise
Linguistic Understanding
Parameter
7B
7B or more
Cantonese Proficiency
Tick Mark Icon
Tick Mark Icon
Multilingual Understanding
Tick Mark Icon
Tick Mark Icon
Domain Knowledge Capability
Tick Mark Icon
Tick Mark Icon
Implementation
Conversation Volume
Tick Mark Icon
Tick Mark Icon
AI Consulting and Engineering Support
Tick Mark Icon
Security & Compliance
Standard
Enterprise-Grade
On-Cloud Support (API) and On-Premise Support (Commercial Use License)
Tick Mark Icon
Custom Model Fine-Tuning
Tick Mark Icon