开源AI网关Aurora推出15MB轻量架构支持多模型自动转换
基于Go语言开发的开源AI网关（AI Gateway）Aurora正式发布。该项目二进制文件体积仅约15MB且无需外部依赖，采用Apache 2.0开源协议，旨在为开发者提供统一的大模型基础设施层服务。
Aurora能够自动将OpenAI SDK的请求格式转换为Anthropic等提供商的协议格式，实现无需修改代码的大模型切换。系统内置多提供商连接池与自动故障转移机制，同时支持语义缓存（Semantic Caching）与精确缓存，有效降低API调用成本与访问延迟。
该工具集成了安全护栏、审计日志导出与可视化管理面板，完善了开源大模型基础设施的监控运维能力。企业版在此基础上进一步提供单点登录与多租户隔离，推动企业级大模型应用的标准化落地。
15MB零依赖架构降低部署门槛
自动协议转换消除厂商绑定
语义缓存与故障转移降本增效
Aurora Launches Open Source AI Gateway 15MB Binary Cross Provider Translation
Aurora announced the official open source availability of its lightweight Artificial Intelligence (AI) gateway software under the Apache 2.0 license. Designed for enterprise software teams, the standalone Go application operates as a compact 15MB binary with zero external runtime dependencies. The platform simplifies infrastructure management by unifying API integration across prominent Large Language Model (LLM) providers, including OpenAI, Anthropic, and Google Gemini, through a centralized control proxy.
The technical framework enables automatic protocol and wire format conversion, allowing backend applications to send OpenAI Software Development Kit (SDK) requests while automatically translating them into Anthropic wire format payloads. Aurora optimizes reliability and cost through provider pooling with automated failover switching, dynamic model auto-discovery, and exact semantic caching. In addition, the system provides detailed cost analytics per customer, built-in security guardrails, compliance audit logging exports, and dedicated administrative endpoints for user onboarding.
As enterprise adoption of multi-provider AI architectures accelerates, managing fragmented SDKs and retry logic creates substantial engineering overhead. Aurora resolves this operational challenge by abstracting provider dependencies into a resilient, high-performance middleware layer. This modern architectural pattern empowers engineering organizations to switch foundation models dynamically, eliminate vendor lock-in, and maintain strict governance across production environment workflows.
Key Takeaways:
Aurora releases a Go-based open source gateway providing multi-provider model aggregation and zero-dependency deployment. Source: Original Article
Automated payload format conversion enables seamless switching between OpenAI and Anthropic API endpoints. Source: Original Article
Built-in semantic caching and provider failover pools significantly reduce enterprise LLM operational costs. Source: Original Article
查看原文 →
View Original →