Developer Builds Vancine, a Cheaper API Focused on Current Chinese AI Models

A developer frustrated with OpenRouter's large, cluttered catalog built Vancine, an OpenAI-compatible API that focuses exclusively on current flagship Chinese language models from providers like Qwen, Kimi, GLM, MiniMax, and DeepSeek. The platform uses a single API key, a single balance, and one endpoint, with a deliberately small catalog that retires older model versions as providers release replacements. As of August 27, 2026, Vancine's pricing on four compared paid listings was approximately 20% lower than OpenRouter's equivalent offerings. Beyond text and coding models, the service also provides access to Chinese image, video, audio, and 3D generation models under the same account. The creator acknowledges the platform is self-run and is still evaluating what trust and transparency signals — such as uptime history and rate-limit details — smaller API providers need to publish for developers to rely on them in production.