Every signal a wave, Every wave an ocean.

We build large language models, AI agents, and standardized corpus libraries for Cantonese, Chinese, and English — engineered in Hong Kong for the Greater Bay Area and the world.

What We Do

Star Ocean operates across four connected business lines. Each strengthens the others, forming a full loop from raw language data to deployed intelligent applications.

Large-Model R&D

Foundation models, tuned for Cantonese, Chinese, and English.

Continued pre-training, fine-tuning, and alignment of large language models for Cantonese, Chinese, and English (yue, zh-Hans, and zh-Hant). We help partners adapt foundation models to their data, domain, and compliance needs.

AI Agent Development

Production-grade agents, end to end.

We design and deliver production-grade AI agents: conversational assistants, workflow automation, and domain-specific copilots tuned for Hong Kong's trilingual users. From prototype to deployment, we own the full engineering lifecycle.

Standardized Corpus Library

Good models start with good data.

We build and curate high-quality, standardized Cantonese and Chinese–English corpora — cleaned, de-duplicated, quality-graded, and license-clear.

Why Star Ocean

Local Teams

Hong Kong–based local R&D teams with on-the-ground business experience across the Greater Bay Area, backed by a dedicated scientist team.

Full-Stack Service

From AI large models to agent building and AI-powered traffic operations — the whole stack, under one roof.

Corpus Library

Cantonese and bilingual corpora built on Hong Kong usage, at 10PB scale.

How We Work

MaaS (Model as Service)

OpenAI API Anthropic API Pay-as-you-go Subscription

Chinese and English large-model access compatible with mainstream OpenAI and Anthropic protocols, offered in both pay-as-you-go and subscription modes. Text LLMs are available today, with ASR models rolling out continuously.

Agent Dev

Custom End-to-End

Customized AI agent development services — from design and integration to deployment and continuous tuning.

Data Governance Service

Cleaning Curation

Corpus cleaning, curation, and standardization at scale — turning raw language data into license-clear, model-ready datasets.

About Star Ocean

Founded in Hong Kong, Star Ocean builds the large-model stack for Cantonese, Chinese, and English — combining deep model R&D and disciplined data engineering to power the next generation of intelligent applications.