alexsun/compose-anything
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: add GoModel

Browse Source
This commit is contained in:
Sun-ZhenXing
2026-05-06 17:14:43 +08:00
parent c59a1e93f6
commit 5e8999c625
6 changed files with 372 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+1
View File
@@ -88,6 +88,7 @@ These services require building custom Docker images from source.
| [Gitea](./src/gitea) | 1.25.4-rootless |
| [GitLab Runner](./src/gitlab-runner) | 17.10.1 |
| [GitLab](./src/gitlab) | 18.8.3-ce.0 |
| [GoModel](./src/gomodel) | v0.1.25 |
| [GPUStack](./src/gpustack) | v0.5.3 |
| [Grafana](./src/grafana) | 12.3.2 |
| [Grafana Loki](./src/loki) | 3.3.2 |
README.zh.md
+1
View File
@@ -88,6 +88,7 @@ docker compose exec redis redis-cli ping
| [Gitea](./src/gitea) | 1.25.4-rootless |
| [GitLab Runner](./src/gitlab-runner) | 17.10.1 |
| [GitLab](./src/gitlab) | 18.8.3-ce.0 |
| [GoModel](./src/gomodel) | v0.1.25 |
| [GPUStack](./src/gpustack) | v0.5.3 |
| [Grafana](./src/grafana) | 12.3.2 |
| [Grafana Loki](./src/loki) | 3.3.2 |
src/gomodel/.env.example
+87
View File
@@ -0,0 +1,87 @@
# GoModel Configuration
# =====================
# Copy this file to .env and fill in the values you need.
# All entries are optional — GoModel starts without any provider key
# configured, but will serve no models until at least one is set.
# ---- Image version ----
GOMODEL_VERSION=0.1.25
# ---- Network ----
# Host port that maps to the container's port 8080
GOMODEL_PORT_OVERRIDE=8080
# ---- Timezone ----
TZ=UTC
# ---- Security ----
# Set a strong secret to protect all API endpoints.
# Without this key the gateway runs in UNSAFE (open) mode.
# Clients must send: Authorization: Bearer <GOMODEL_MASTER_KEY>
# GOMODEL_MASTER_KEY=change-me-before-exposing-to-network
# ---- Storage ----
# Backend for audit logs and usage data.
# Options: sqlite (default), postgresql, mongodb
GOMODEL_STORAGE_TYPE=sqlite
# SQLite database file path inside the container (persisted via named volume)
# SQLITE_PATH=/app/data/gomodel.db
# ---- Audit logging ----
# Log all requests and responses to the configured storage backend.
# WARNING: May capture PII and API key values present in prompts.
GOMODEL_LOGGING_ENABLED=false
# ---- Resource limits ----
GOMODEL_CPU_LIMIT=1.00
GOMODEL_MEMORY_LIMIT=512M
GOMODEL_CPU_RESERVATION=0.25
GOMODEL_MEMORY_RESERVATION=128M
# ===========================================================================
# Provider API Keys
# Set the credentials for every provider you want to route traffic through.
# Providers without a key (or base URL) are simply not registered.
# ===========================================================================
# OpenAI
# OPENAI_API_KEY=sk-...
# OPENAI_BASE_URL=https://api.openai.com/v1
# Anthropic
# ANTHROPIC_API_KEY=sk-ant-...
# Google Gemini
# GEMINI_API_KEY=...
# Groq
# GROQ_API_KEY=gsk_...
# xAI (Grok)
# XAI_API_KEY=...
# DeepSeek
# DEEPSEEK_API_KEY=...
# OpenRouter
# OPENROUTER_API_KEY=sk-or-...
# Z.ai
# ZAI_API_KEY=...
# Azure OpenAI
# AZURE_API_KEY=...
# AZURE_BASE_URL=https://your-resource.openai.azure.com/openai/deployments/your-deployment
# AZURE_API_VERSION=2024-10-21
# Oracle
# ORACLE_API_KEY=...
# ORACLE_BASE_URL=https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/20231130/actions/v1
# ORACLE_MODELS=openai.gpt-oss-120b,xai.grok-3
# Ollama (local inference server — no API key needed for default setup)
# OLLAMA_BASE_URL=http://host.docker.internal:11434/v1
# vLLM (OpenAI-compatible server — API key optional)
# VLLM_BASE_URL=http://host.docker.internal:8000/v1
# VLLM_API_KEY=token-abc123
src/gomodel/README.md
+110
View File
@@ -0,0 +1,110 @@
# GoModel
[GoModel](https://github.com/ENTERPILOT/GoModel) is a fast, lightweight AI gateway written in Go. It exposes a single unified OpenAI-compatible API that transparently routes requests to OpenAI, Anthropic, Gemini, Groq, xAI, DeepSeek, OpenRouter, Azure OpenAI, Oracle, Ollama, vLLM, and more. It ships a built-in admin dashboard with token usage tracking, cost estimation, audit logging, and response caching.
## Services
| Service | Port | Description |
| ------- | ---- | ---------------------------------------------------- |
| GoModel | 8080 | OpenAI-compatible AI gateway API and admin dashboard |
## Quick Start
```bash
docker compose up -d
```
The gateway is available at `http://localhost:8080`. By default it starts in open (unauthenticated) mode with SQLite storage and no provider keys configured.
To route real traffic, create a `.env` file from the example and add at least one provider credential:
```bash
cp .env.example .env
# Edit .env — set GOMODEL_MASTER_KEY and at least one provider API key
docker compose up -d
```
Make your first API call:
```bash
curl http://localhost:8080/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <GOMODEL_MASTER_KEY>" \
-d '{
"model": "gpt-4o",
"messages": [{"role": "user", "content": "Hello!"}]
}'
```
Admin dashboard: `http://localhost:8080/admin/dashboard`
List available models: `http://localhost:8080/v1/models`
## Key Environment Variables
### Gateway
| Variable | Default | Description |
| ------------------------- | --------- | ------------------------------------------------------------------ |
| `GOMODEL_VERSION` | `0.1.25` | Docker image version |
| `GOMODEL_PORT_OVERRIDE` | `8080` | Host port for the API |
| `GOMODEL_MASTER_KEY` | _(empty)_ | API authentication key — **set this for any non-local deployment** |
| `GOMODEL_STORAGE_TYPE` | `sqlite` | Storage backend: `sqlite`, `postgresql`, or `mongodb` |
| `GOMODEL_LOGGING_ENABLED` | `false` | Enable full audit logging of requests and responses |
| `TZ` | `UTC` | Container timezone |
### Provider Credentials
Set the keys for whichever providers you want to use. Unset entries are silently ignored.
| Variable | Provider |
| ------------------------------------ | ----------------------------- |
| `OPENAI_API_KEY` | OpenAI |
| `ANTHROPIC_API_KEY` | Anthropic |
| `GEMINI_API_KEY` | Google Gemini |
| `GROQ_API_KEY` | Groq |
| `XAI_API_KEY` | xAI (Grok) |
| `DEEPSEEK_API_KEY` | DeepSeek |
| `OPENROUTER_API_KEY` | OpenRouter |
| `ZAI_API_KEY` | Z.ai |
| `AZURE_API_KEY` + `AZURE_BASE_URL` | Azure OpenAI |
| `ORACLE_API_KEY` + `ORACLE_BASE_URL` | Oracle |
| `OLLAMA_BASE_URL` | Ollama (local, no key needed) |
| `VLLM_BASE_URL` | vLLM (local, key optional) |
See `.env.example` for the full list including Azure, Oracle, and per-provider model lists.
## Storage
| Volume | Mount | Description |
| --------------- | ------------- | ------------------------------------------ |
| `gomodel_data` | `/app/data` | SQLite database, usage, and audit log data |
| `gomodel_cache` | `/app/.cache` | Model metadata registry cache |
The default storage backend is SQLite (`/app/data/gomodel.db`). Switch to PostgreSQL or MongoDB by setting `GOMODEL_STORAGE_TYPE` and the corresponding `POSTGRES_URL` or `MONGODB_URL` environment variable.
## API Endpoints
| Endpoint | Description |
| --------------------------- | --------------------------------------------------- |
| `GET /v1/models` | List available models from all configured providers |
| `POST /v1/chat/completions` | Chat completions (streaming supported) |
| `POST /v1/embeddings` | Text embeddings |
| `POST /v1/responses` | OpenAI Responses API |
| `GET /health` | Health probe |
| `GET /admin/dashboard` | Admin UI — usage, costs, audit log |
| `GET /metrics` | Prometheus metrics (when `METRICS_ENABLED=true`) |
## Security Notes
- **`GOMODEL_MASTER_KEY` is unset by default.** Without it, all API endpoints are publicly accessible. Set a strong secret before exposing the service on any network beyond localhost.
- The container runs as a non-root user (UID 65532) on a read-only root filesystem.
- `cap_drop: [ALL]` — no Linux capabilities required.
- Never pass API keys via the command line; use the `.env` file or environment variables.
- Audit log bodies (`LOGGING_LOG_BODIES`) may capture PII and API keys embedded in prompts — keep it disabled unless you control the data retention pipeline.
## Links
- [GitHub Repository](https://github.com/ENTERPILOT/GoModel)
- [Documentation](https://gomodel.enterpilot.io/docs)
- [Docker Hub](https://hub.docker.com/r/enterpilot/gomodel)
src/gomodel/README.zh.md
+110
View File
@@ -0,0 +1,110 @@
# GoModel
[GoModel](https://github.com/ENTERPILOT/GoModel) 是一个用 Go 编写的快速、轻量级 AI 网关。它通过单一统一的 OpenAI 兼容 API，透明地将请求路由到 OpenAI、Anthropic、Gemini、Groq、xAI、DeepSeek、OpenRouter、Azure OpenAI、Oracle、Ollama、vLLM 等众多提供商。内置管理仪表盘，支持 Token 用量追踪、成本估算、审计日志和响应缓存。
## 服务组件
| 服务 | 端口 | 说明 |
| ------- | ---- | -------------------------------------- |
| GoModel | 8080 | OpenAI 兼容的 AI 网关 API 及管理仪表盘 |
## 快速开始
```bash
docker compose up -d
```
服务启动后可通过 `http://localhost:8080` 访问。默认以开放（无鉴权）模式运行，使用 SQLite 存储，且未配置任何提供商密钥。
接入真实流量，请先从示例文件创建 `.env` 并填入至少一个提供商凭据：
```bash
cp .env.example .env
# 编辑 .env，设置 GOMODEL_MASTER_KEY 以及至少一个提供商 API Key
docker compose up -d
```
发起第一次 API 调用：
```bash
curl http://localhost:8080/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <GOMODEL_MASTER_KEY>" \
-d '{
"model": "gpt-4o",
"messages": [{"role": "user", "content": "你好！"}]
}'
```
管理仪表盘：`http://localhost:8080/admin/dashboard`
查看可用模型列表：`http://localhost:8080/v1/models`
## 关键环境变量
### 网关配置
| 变量 | 默认值 | 说明 |
| ------------------------- | -------- | ------------------------------------------------ |
| `GOMODEL_VERSION` | `0.1.25` | Docker 镜像版本 |
| `GOMODEL_PORT_OVERRIDE` | `8080` | API 宿主机端口 |
| `GOMODEL_MASTER_KEY` | _（空）_ | API 鉴权密钥——**向本机以外的网络暴露时必须设置** |
| `GOMODEL_STORAGE_TYPE` | `sqlite` | 存储后端：`sqlite`、`postgresql` 或 `mongodb` |
| `GOMODEL_LOGGING_ENABLED` | `false` | 是否启用完整的请求/响应审计日志 |
| `TZ` | `UTC` | 容器时区 |
### 提供商凭据
仅需设置你要使用的提供商密钥，未设置的条目会被静默忽略。
| 变量 | 提供商 |
| ------------------------------------ | ------------------------ |
| `OPENAI_API_KEY` | OpenAI |
| `ANTHROPIC_API_KEY` | Anthropic |
| `GEMINI_API_KEY` | Google Gemini |
| `GROQ_API_KEY` | Groq |
| `XAI_API_KEY` | xAI（Grok） |
| `DEEPSEEK_API_KEY` | DeepSeek |
| `OPENROUTER_API_KEY` | OpenRouter |
| `ZAI_API_KEY` | Z.ai |
| `AZURE_API_KEY` + `AZURE_BASE_URL` | Azure OpenAI |
| `ORACLE_API_KEY` + `ORACLE_BASE_URL` | Oracle |
| `OLLAMA_BASE_URL` | Ollama（本地，无需密钥） |
| `VLLM_BASE_URL` | vLLM（本地，密钥可选） |
完整变量列表（含 Azure、Oracle 及各提供商模型列表配置）请参见 `.env.example`。
## 存储
| 数据卷 | 挂载路径 | 说明 |
| --------------- | ------------- | --------------------------------- |
| `gomodel_data` | `/app/data` | SQLite 数据库、用量数据及审计日志 |
| `gomodel_cache` | `/app/.cache` | 模型元数据注册表缓存 |
默认使用 SQLite 存储（`/app/data/gomodel.db`）。如需切换至 PostgreSQL 或 MongoDB，请设置 `GOMODEL_STORAGE_TYPE` 及对应的 `POSTGRES_URL` 或 `MONGODB_URL` 环境变量。
## API 端点
| 端点 | 说明 |
| --------------------------- | ------------------------------------------------ |
| `GET /v1/models` | 列出所有已配置提供商的可用模型 |
| `POST /v1/chat/completions` | 聊天补全（支持流式输出） |
| `POST /v1/embeddings` | 文本向量嵌入 |
| `POST /v1/responses` | OpenAI Responses API |
| `GET /health` | 健康探针 |
| `GET /admin/dashboard` | 管理界面——用量、成本、审计日志 |
| `GET /metrics` | Prometheus 指标（需设置 `METRICS_ENABLED=true`） |
## 安全说明
- **`GOMODEL_MASTER_KEY` 默认为空。** 未设置时，所有 API 端点均可公开访问。在将服务暴露到 localhost 以外的网络前，请务必设置强密钥。
- 容器以非 root 用户（UID 65532）运行，根文件系统为只读。
- `cap_drop: [ALL]`——无需任何 Linux 特权能力。
- 请勿通过命令行传递 API 密钥；应使用 `.env` 文件或环境变量。
- 审计日志体（`LOGGING_LOG_BODIES`）可能记录提示词中的敏感信息和 API 密钥，请在具备完整数据保留管控的情况下再启用。
## 相关链接
- [GitHub 仓库](https://github.com/ENTERPILOT/GoModel)
- [官方文档](https://gomodel.enterpilot.io/docs)
- [Docker Hub](https://hub.docker.com/r/enterpilot/gomodel)
src/gomodel/docker-compose.yaml
+63
View File
@@ -0,0 +1,63 @@
x-defaults: &defaults
restart: unless-stopped
logging:
driver: json-file
options:
max-size: 100m
max-file: '3'
services:
gomodel:
<<: *defaults
image: ${GLOBAL_REGISTRY:-}enterpilot/gomodel:${GOMODEL_VERSION:-0.1.25}
ports:
- '${GOMODEL_PORT_OVERRIDE:-8080}:8080'
volumes:
- gomodel_data:/app/data
- gomodel_cache:/app/.cache
read_only: true
cap_drop:
- ALL
environment:
- TZ=${TZ:-UTC}
# Storage backend: sqlite (default), postgresql, or mongodb
- STORAGE_TYPE=${GOMODEL_STORAGE_TYPE:-sqlite}
- SQLITE_PATH=/app/data/gomodel.db
# Master key for API authentication — strongly recommended for production;
# leave empty to run in unsafe/open mode (local evaluation only)
- GOMODEL_MASTER_KEY=${GOMODEL_MASTER_KEY:-}
# Enable full audit logging of requests and responses to storage
- LOGGING_ENABLED=${GOMODEL_LOGGING_ENABLED:-false}
# Provider API keys — set at least one; unused entries are safely ignored
- OPENAI_API_KEY=${OPENAI_API_KEY:-}
- ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY:-}
- GEMINI_API_KEY=${GEMINI_API_KEY:-}
- GROQ_API_KEY=${GROQ_API_KEY:-}
- XAI_API_KEY=${XAI_API_KEY:-}
- DEEPSEEK_API_KEY=${DEEPSEEK_API_KEY:-}
- OPENROUTER_API_KEY=${OPENROUTER_API_KEY:-}
- ZAI_API_KEY=${ZAI_API_KEY:-}
# Ollama / vLLM base URLs (no key required for default local setups)
- OLLAMA_BASE_URL=${OLLAMA_BASE_URL:-}
- VLLM_BASE_URL=${VLLM_BASE_URL:-}
# The image is built on distroless/static-debian12:nonroot which ships no
# shell, wget, or curl. /gomodel --version exits 0 while the binary is
# alive and serves as a process-level liveness probe.
healthcheck:
test: [CMD, /gomodel, --version]
interval: 30s
timeout: 5s
retries: 3
start_period: 15s
deploy:
resources:
limits:
cpus: ${GOMODEL_CPU_LIMIT:-1.00}
memory: ${GOMODEL_MEMORY_LIMIT:-512M}
reservations:
cpus: ${GOMODEL_CPU_RESERVATION:-0.25}
memory: ${GOMODEL_MEMORY_RESERVATION:-128M}
volumes:
gomodel_data:
gomodel_cache: