LiteLLM

##Introduction LiteLLM is a Python SDK and proxy server (LLM Gateway) developed by BerriAI. It can uniformly call more than 100 mainstream large language models (such as Bedrock, Azure, OpenAI, VertexAI, Cohere, Anthropic, HuggingFace, Replicate, etc.), and follows OpenAI's input and output format throughout the process, greatly simplifying the multi-platform access process.  The proxy server part of LiteLLM is currently ported. ##Core Features - Unified interface: Translate user calls such as completion, embedding, image_generation, etc. to APIs of different providers; - Consistent output: Ensure that all text responses are located in choices[0].message.content; - Retries and downgrades: Configure automatic retry, fault fallback, and routing policies across vendors; - Budget and current limits: Flexible setting of budget and rate limits by project, API Key or model; - Proxy services: provide authentication hooks, logging, cost statistics and rate control; - Visual management: Built-in Web UI, online generation and management of proxy keys.  ##Migration Description - LiteLLM mirror version is https://github.com/berriai/litellm/pkgs/container/litellm/418776662? tag=main-v1.70.1-stable - https://litellm. <lzc-name>. heiyu.space/ui You can enter the management page, default user name: admin, default password: sk-1234