AI platforms
AstraFlow星图
One API Key to access 200+ leading LLMs with seamless OpenAI protocol compatibil
What is AstraFlow星图?
AstraFlow is a large model aggregation service platform launched by UCloud, enabling direct access to over 200 mainstream large models including DeepSeek, Qwen, GLM, Kimi, and MiniMax through a single API Key.It covers full-stack AI capabilities across language, image, and video domains, supporting scenarios such as dialogue, reasoning, code generation, text-to-image, and video generation to meet diverse development needs in one place.
Fully compatible with the OpenAI API protocol, AstraFlow allows developers to integrate without modifying existing code—simply replace the base_url and API Key to get started in minutes.The platform ensures hour-level synchronization for the latest popular models and provides enterprise-grade security and compliance on UCloud's infrastructure, guaranteeing that user data is never used for model training while offering competitive pricing without intermediate markups.
Key features
- Massive Model AgAggregates 200+ mainstream large models from 15 vendors including DeepSeek, Qwen, Zhipu GLM, Kimi, MiniMax, and Doubao, covering language, image, and video domains.
- OpenAI ProtocolFully compatible with the OpenAI API protocol, supporting mainstream frameworks like OpenAI Python SDK and LangChain, allowing seamless migration and invocation without code changes.
- Hour-Level ModelMaintains close cooperation with major model vendors, ensuring hour-level synchronization for new popular models like DeepSeek-V4 and Qwen3 upon release.
- Full-Stack AI CaSupports not only text dialogue and reasoning but also integrates text-to-image, image-to-image, visual understanding, and text/video-to-video multimodal generation.
- Enterprise-GradeBuilt on UCloud cloud infrastructure with Level 3 security certification, promising never to use user data for model training, and supporting private deployment and VPC internal network calls.
- Competitive PricDirect official connection removes intermediate markups, offering original factory discounts with pay-as-you-go billing, no minimum consumption, and peak/valley pricing for some models to reduce costs.
How to use it
- Register for a UCloud account and complete real-name verification to instantly unlock access to 200+ mainstream large models.
- Log in to the UCloud console, navigate to the AstraFlow model service platform, and create your exclusive API Key with one click.
- Configure the base_url to https://api.modelverse.cn in your code or AI tool, and insert the generated API Key.
- Specify the model ID you wish to call (e.g., qwen-max or deepseek-v3) to start sending requests and receiving responses.
Where it fits
- Rapid ApplicatioApplications already developed based on the OpenAI interface can switch to AstraFlow by modifying just two lines of configuration, enabling low-cost access to mainstream domestic large models.
- Multimodal ConteLeverage integrated text-to-image and text-to-video models to quickly generate marketing materials, creative videos, and multi-style images, boosting content production efficiency.
- Enterprise IntelCombine long-context and cost-effective reasoning models to build intelligent customer service systems capable of handling complex logic and long-term memory, reducing operational costs.
Pricing and free quota
Pay-as-you-go pricing model with varying rates per model and no minimum consumption.For example, DeepSeek-V4.1-Flash-1017 is ¥2/M for input during peak hours and ¥1/M during off-peak hours.Please refer to the official website for real-time pricing.
Who it is for
- AI Application Developers
- Enterprise Technical Teams
- Content Creators
- Independent Developers
FAQ
Which large models does AstraFlow support?
The platform has integrated 200+ large models from 15 vendors including Baidu Wenxin, ByteDance, DeepSeek, MiniMax, Moonshot AI, Qwen, and Zhipu AI, covering language, image, and video generation.
How is data security and privacy protected?
Built on UCloud's infrastructure with Level 3 security certification, data transmission is fully encrypted.The platform promises not to store user request content or use it for model training, with private deployment available for enterprise editions.
How soon can new models be called after release?
AstraFlow maintains close cooperation with model vendors, achieving hour-level synchronization for mainstream large models upon release.Developers can query the list of currently available models via the API's models interface.