fuzhengwei/waliapi
Developer toolsThis page was translated automatically. Technical details may differ from the original. View the original
WaLiaPI is a locally run LLM API gateway desktop software (Tauri 2 + React + Rust). Support for access to various LLM manufacturers, Ollama self-deployment model, CPA approach, ChatGPT access, and unified support for the conversion of Chat Corporations / Reponses / Anthropic Messages。
Overview
🎁: Start-up is built-in with two free Key channels, zero configuration direct call; and provide - International (Groq / Cerebras / Mistral / NVIDIA NIM / OpenRouter / GitHub Models / Hugging Face / SambaNova / Cohere) and Domestic (ModelScope/ SiliconFlow Silicon-based Movement) - Registered for API Key, which is also available at Google Gemini and GLM. For more details on the free channels. This chart below describes the full workflow of WaLiaPI: the left side is the AI application for the request, with the middle being the processing link for the WaLiaPI gateway (with the built-in RAG knowledge base and LLM-Wiki knowledge engine) and the right side is the three upstream channels you have configured - (Key access), (free Key/ free of charge) and (Auth account number)** (OAuth login). : All downstream AI applications point to WaLAIAPI - it completes protocol translation, security audit, load balance and log audit and forwards the request to the API channel, free channel or account number channel that you have configured; automatically switch to the next AI tool when a certain channel is exhausted or malfunctioning. The gateway incorporates the RAG knowledge base and LLM-Wiki knowledge engine, where Agent can directly access private knowledge through MCP/REST, and the model of the knowledge service calls for the same channels of movement and auditing. Seed channels are added only once and are not automatically added when deleted; you can edit, start or participate in the scheduling like normal channels. In the New Channel, select the following preset to register on the interface 'API Key 'link, and the model list supports "Synthetic Upstream Model" withdrawal (Autofiltration Free Available Model): In addition, the preset description of Google Gemini (AI Studio Free Level, Flash Series 1 million token, 15 RPM) and the GLM (gram-4.7-flash etc. Free of charge for Flash Series) has been marked free layer description and registration entrance. • Free-of-charge access levels are subject to a different rate, with an upstream return of 429, overtime, and the gateway will automatically switch to the next available channel. In addition, account lines (e.g. Antigravity, Codex, Kimi, Qoder) may be used by the "Auth Account" OAuth login, with some suppliers providing a free or subscription line。
Overview (machine translated)
Local LLM API gateway # Multi Protocol access # Knowledge base RG MCP Tool service
Unify upstream model suppliers such as OpenAI, Claude, DeepSeek and Gemini as a local gateway, ** Access to AI programming tools such as Claude Code, Codex CLI, Gemini CLI, WaLiCode, etc. with a set of keys, a set of protocols, a full audit log. **
** Download installation** & nbsp; & nbsp; ** Core function** & nbsp; & nbsp; ** Rationale** & nbsp; & nbsp; ** Docker deployment**
** C Chinese** & nbsp; English & nbsp; |nbsp; Japanese
** Free channel opening is : Inline: ** OpenCode Free Drive and Kilo Free Pool with two free Key Channels** and zero configurations; and **11 Free Channel preset ** - International (Groq / Cerebras / Middle / NVIDIA Nim / OpenRouter / GitHub Models / Hugging Face / Sambanova / Cohere) and National (ModelScope/ SiliconFlow Silicon-based) - also registered for API Key, which is also available free of charge in Google Gemini and GLMM. For more details on the free channels.
I don’t…
givin’ 30 seconds to see WaLiaPI
The following figure describes the full workflow of WaLiaPI: the AI application for the request on the left, the processing link for the WaLiaPI gateway (with the built-in RAG knowledge base and LLM-Wiki knowledge engine) and the API channel** (Key access),** free of charge** and account number** (Auth account)** on the right.
** Summary of sentence**: All downstream AI applications point to WaLiaPI - which completes protocol translation, security audit, load balance and log audit and forwards the request to the API channel, free channel or account number channel that you have configured; and automatically switch to the next AI tool when a certain channel is exhausted or malfunctioning. The gateway incorporates the RAG knowledge base and LLM-Wiki knowledge engine, where Agent can directly access private knowledge through MCP/REST, and the model of the knowledge service calls for the same channels of movement and auditing.
It’s free to open.
You can run without a penny. WaLiaPI provides two types of free channels:
1. Leave Key built-in channel (zero configuration)
Automatically pre-engineer the “channel management” on the first start, without registration, without Key, and with key saver:
Information on channels |: -----------
This page was translated automatically. Technical details may differ from the original.
Install
docker run -d --name waliapi \