# yaml-language-server: $schema=https://schema.zeabur.app/template.json apiVersion: zeabur.com/v1 kind: Template metadata: name: Bifrost spec: description: Self-hosted Bifrost AI gateway — OpenAI-compatible API plus dashboard, pinned to v2.0.0, SQLite, no extra services. coverImage: https://opengraph.githubassets.com/1/maximhq/bifrost icon: https://avatars.githubusercontent.com/maximhq variables: - key: PUBLIC_DOMAIN type: DOMAIN name: Gateway Domain description: Domain for the Bifrost dashboard and OpenAI-compatible API. - key: OPENAI_API_KEY type: STRING name: OpenAI API Key (Optional) description: Optional. Upstream OpenAI key. Leave blank and add it later in the dashboard. - key: ANTHROPIC_API_KEY type: STRING name: Anthropic API Key (Optional) description: Optional. Upstream Anthropic key. Leave blank to skip. - key: OPENROUTER_API_KEY type: STRING name: OpenRouter API Key (Optional) description: Optional. Upstream OpenRouter key. Leave blank to skip. - key: DEEPSEEK_API_KEY type: STRING name: DeepSeek API Key (Optional) description: Optional. Upstream DeepSeek key. Leave blank to skip. tags: - Tool - AI - Gateway - LLM readme: |- # Bifrost [Bifrost](https://github.com/maximhq/bifrost) is an open-source AI gateway. This template follows the official [runtime contract](https://docs.getbifrost.ai/deployment-guides/runtime-contract) (image `v2.0.0`): one Go process, SQLite on `/app/data`, dashboard and inference on the same HTTP port. The stack: - **bifrost** — `maximhq/bifrost:v2.0.0` (API, dashboard, SQLite config store, SQLite logs store) No PostgreSQL, Redis, or extra workers. OSS SQLite is single-replica only. ## Resource recommendation The image is ~87 MB. 0.5–1 vCPU and 512 MB–1 GB RAM is enough for personal or small-team traffic. ## Configuration The gateway domain is required. Dashboard auth is on from first boot (`admin` / auto-generated password). Find `BIFROST_ADMIN_PASSWORD` and `BIFROST_VIRTUAL_KEY` on the bifrost service env tab. Optional upstream keys (`OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, `OPENROUTER_API_KEY`, `DEEPSEEK_API_KEY`) are wired through `config.json` as `env.*` references. Leave them blank and add providers in the dashboard if you prefer. More providers (Gemini, Azure, Bedrock, Groq, Ollama, …) can be added in the UI without redeploying. `BIFROST_ENCRYPTION_KEY` is auto-generated and must stay stable across restarts. Changing it makes persisted secrets unreadable. ## First login Open the gateway domain. Sign in with: - username: `admin` - password: `BIFROST_ADMIN_PASSWORD` from the env tab Then create extra virtual keys, budgets, or providers in the dashboard. ## Quick test After `/health` returns 200 and at least one upstream key is set: ```bash curl --fail-with-body --max-time 60 \ -X POST https:///v1/chat/completions \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{"model":"openai/gpt-4o-mini","messages":[{"role":"user","content":"Hello"}]}' ``` Admin basic auth also works: `-u admin:`. Drop-in OpenAI SDK / GapSeek `OPENAI_BASE_URL`: ```text https:///openai ``` Use `BIFROST_VIRTUAL_KEY` as `OPENAI_API_KEY`. Model ids are `provider/model`, e.g. `openai/gpt-4o-mini`, `anthropic/claude-sonnet-4-5`, `openrouter/deepseek/deepseek-chat`. ## What this template includes | Need | This template | | --- | --- | | OpenAI-compatible `/v1` + `/openai` | Included | | Built-in dashboard | Included (same domain) | | Failover / load balance / virtual keys | Included (configure in UI) | | Semantic cache, MCP, extra providers | Add in the dashboard | | Multi-replica / HA | Not included; needs Bifrost Enterprise + PostgreSQL | Do not expose an unauthenticated gateway. Inference calls require the virtual key (or admin basic auth) unless you turn that off in Security settings. ## Versioning Image is pinned to `maximhq/bifrost:v2.0.0` by digest (pulled 2026-09-03). Upstream also publishes `:latest`. ## Reference - Setup: https://docs.getbifrost.ai/quickstart/gateway/setting-up - Runtime contract: https://docs.getbifrost.ai/deployment-guides/runtime-contract - config.json: https://docs.getbifrost.ai/deployment-guides/config-json - GitHub: https://github.com/maximhq/bifrost - License: Apache-2.0 services: - name: bifrost icon: https://avatars.githubusercontent.com/maximhq template: PREBUILT spec: id: bifrost source: image: maximhq/bifrost:v2.0.0@sha256:cf71be9fad4e0749b6e26cbb774c687413dad9a0970b83f4e1dadb6f503ea208 ports: - id: web port: 8080 type: HTTP volumes: - id: data dir: /app/data instructions: - title: Bifrost URL content: ${ZEABUR_WEB_URL} - title: Dashboard content: ${ZEABUR_WEB_URL} - title: Admin username content: ${BIFROST_ADMIN_USERNAME} - title: Admin password content: ${BIFROST_ADMIN_PASSWORD} - title: Virtual key (Bearer for /v1) content: ${BIFROST_VIRTUAL_KEY} - title: OpenAI-compatible base URL content: ${ZEABUR_WEB_URL}/openai - title: Health content: ${ZEABUR_WEB_URL}/health - title: Quick chat (curl) content: |- curl --fail-with-body --max-time 60 \ -X POST ${ZEABUR_WEB_URL}/v1/chat/completions \ -H "Authorization: Bearer ${BIFROST_VIRTUAL_KEY}" \ -H "Content-Type: application/json" \ -d '{"model":"openai/gpt-4o-mini","messages":[{"role":"user","content":"Hello"}]}' env: ANTHROPIC_API_KEY: default: ${ANTHROPIC_API_KEY} APP_DIR: default: /app/data APP_HOST: default: 0.0.0.0 APP_PORT: default: "8080" BIFROST_ADMIN_PASSWORD: default: Bifrost-${PASSWORD}-Aa1! expose: true BIFROST_ADMIN_USERNAME: default: admin expose: true BIFROST_ENCRYPTION_KEY: default: ${PASSWORD} BIFROST_SETUP_TOKEN: default: ${PASSWORD} BIFROST_URL: default: ${ZEABUR_WEB_URL} expose: true BIFROST_VIRTUAL_KEY: default: vk-${PASSWORD} expose: true DEEPSEEK_API_KEY: default: ${DEEPSEEK_API_KEY} LOG_LEVEL: default: info LOG_STYLE: default: json OPENAI_API_KEY: default: ${OPENAI_API_KEY} OPENROUTER_API_KEY: default: ${OPENROUTER_API_KEY} configs: - path: /app/data/config.json template: |- { "$schema": "https://www.getbifrost.ai/schema", "encryption_key": "env.BIFROST_ENCRYPTION_KEY", "setup_token": "env.BIFROST_SETUP_TOKEN", "client": { "drop_excess_requests": false, "enable_logging": true, "enforce_auth_on_inference": true }, "governance": { "auth_config": { "is_enabled": true, "admin_username": "env.BIFROST_ADMIN_USERNAME", "admin_password": "env.BIFROST_ADMIN_PASSWORD", "disable_auth_on_inference": false }, "virtual_keys": [ { "id": "vk-default", "name": "default", "value": "env.BIFROST_VIRTUAL_KEY", "is_active": true, "provider_configs": [ { "provider": "openai", "allowed_models": ["*"], "key_ids": ["*"], "weight": 1 }, { "provider": "anthropic", "allowed_models": ["*"], "key_ids": ["*"], "weight": 1 }, { "provider": "openrouter", "allowed_models": ["*"], "key_ids": ["*"], "weight": 1 }, { "provider": "deepseek", "allowed_models": ["*"], "key_ids": ["*"], "weight": 1 } ] } ] }, "providers": { "openai": { "keys": [ { "name": "openai-primary", "value": "env.OPENAI_API_KEY", "models": ["*"], "weight": 1.0 } ] }, "anthropic": { "keys": [ { "name": "anthropic-primary", "value": "env.ANTHROPIC_API_KEY", "models": ["*"], "weight": 1.0 } ] }, "openrouter": { "keys": [ { "name": "openrouter-primary", "value": "env.OPENROUTER_API_KEY", "models": ["*"], "weight": 1.0 } ] }, "deepseek": { "keys": [ { "name": "deepseek-primary", "value": "env.DEEPSEEK_API_KEY", "models": ["*"], "weight": 1.0 } ] } }, "config_store": { "enabled": true, "type": "sqlite", "config": { "path": "./config.db" } }, "logs_store": { "enabled": true, "type": "sqlite", "config": { "path": "./logs.db" } } } permission: null envsubst: false healthCheck: type: HTTP port: web http: path: /health domainKey: PUBLIC_DOMAIN localization: zh-CN: description: 自托管 Bifrost AI 网关 — OpenAI 兼容 API + 控制台，锁定 v2.0.0，SQLite，无额外服务。 variables: - key: PUBLIC_DOMAIN type: DOMAIN name: 网关域名 description: Bifrost 控制台和 OpenAI 兼容 API 要绑定的域名。 - key: OPENAI_API_KEY type: STRING name: OpenAI API Key（可选） description: 可选。上游 OpenAI key。留空可稍后在控制台添加。 - key: ANTHROPIC_API_KEY type: STRING name: Anthropic API Key（可选） description: 可选。上游 Anthropic key。留空则跳过。 - key: OPENROUTER_API_KEY type: STRING name: OpenRouter API Key（可选） description: 可选。上游 OpenRouter key。留空则跳过。 - key: DEEPSEEK_API_KEY type: STRING name: DeepSeek API Key（可选） description: 可选。上游 DeepSeek key。留空则跳过。 readme: |- # Bifrost [Bifrost](https://github.com/maximhq/bifrost) 是开源 AI 网关。本模板按官方 [runtime contract](https://docs.getbifrost.ai/deployment-guides/runtime-contract)（镜像 `v2.0.0`）部署：单个 Go 进程、SQLite 落在 `/app/data`，控制台和推理共用同一个 HTTP 端口。 包含服务： - **bifrost** — `maximhq/bifrost:v2.0.0`（API、控制台、SQLite 配置库、SQLite 日志库） 不需要 PostgreSQL、Redis 或额外 worker。OSS 的 SQLite 只支持单副本。 ## 资源建议 镜像约 87 MB。个人或小团队 0.5–1 vCPU、512 MB–1 GB RAM 即可。 ## 配置 需要填写网关域名。控制台鉴权开箱即开（用户名 `admin`，密码自动生成）。到 bifrost 服务的 env 里查看 `BIFROST_ADMIN_PASSWORD` 和 `BIFROST_VIRTUAL_KEY`。 可选上游 key（`OPENAI_API_KEY`、`ANTHROPIC_API_KEY`、`OPENROUTER_API_KEY`、`DEEPSEEK_API_KEY`）通过 `config.json` 的 `env.*` 引用注入。留空也可以，之后在控制台补。Gemini、Azure、Bedrock、Groq、Ollama 等可在 UI 里加，不用重新部署。 `BIFROST_ENCRYPTION_KEY` 会自动生成，重启后不要改。改了会导致已加密的配置读不出来。 ## 首次登录 打开网关域名，使用： - 用户名：`admin` - 密码：env 里的 `BIFROST_ADMIN_PASSWORD` 然后可在控制台继续加虚拟 Key、预算或其它供应商。 ## 快速测试 `/health` 返回 200，且至少配置了一个上游 key 之后： ```bash curl --fail-with-body --max-time 60 \ -X POST https:///v1/chat/completions \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{"model":"openai/gpt-4o-mini","messages":[{"role":"user","content":"Hello"}]}' ``` 管理员 Basic Auth 也可以：`-u admin:`。 对接 OpenAI SDK / GapSeek 的 `OPENAI_BASE_URL`： ```text https:///openai ``` `OPENAI_API_KEY` 填 `BIFROST_VIRTUAL_KEY`。模型名是 `provider/model`，例如 `openai/gpt-4o-mini`、`anthropic/claude-sonnet-4-5`、`openrouter/deepseek/deepseek-chat`。 ## 本模板范围 | 需求 | 本模板 | | --- | --- | | OpenAI 兼容 `/v1` + `/openai` | 已包含 | | 内置控制台 | 已包含（同一域名） | | 故障转移 / 负载均衡 / 虚拟 Key | 已包含（在 UI 配置） | | 语义缓存、MCP、更多供应商 | 在控制台添加 | | 多副本 / HA | 不包含；需要 Bifrost Enterprise + PostgreSQL | 不要把未鉴权的网关暴露到公网。推理请求默认需要虚拟 Key（或管理员 Basic Auth），除非你在 Security 里关掉 inference 鉴权。 ## 版本 镜像按 digest 锁定 `maximhq/bifrost:v2.0.0`（拉取日期 2026-09-03）。 ## 参考 - 安装: https://docs.getbifrost.ai/quickstart/gateway/setting-up - Runtime contract: https://docs.getbifrost.ai/deployment-guides/runtime-contract - config.json: https://docs.getbifrost.ai/deployment-guides/config-json - GitHub: https://github.com/maximhq/bifrost - License: Apache-2.0 zh-TW: description: 自架 Bifrost AI 閘道 — OpenAI 相容 API + 控制台，鎖定 v2.0.0，SQLite，無額外服務。 variables: - key: PUBLIC_DOMAIN type: DOMAIN name: 閘道網域 description: Bifrost 控制台與 OpenAI 相容 API 要綁定的網域。 - key: OPENAI_API_KEY type: STRING name: OpenAI API Key（選填） description: 選填。上游 OpenAI key。留空可稍後在控制台新增。 - key: ANTHROPIC_API_KEY type: STRING name: Anthropic API Key（選填） description: 選填。上游 Anthropic key。留空則略過。 - key: OPENROUTER_API_KEY type: STRING name: OpenRouter API Key（選填） description: 選填。上游 OpenRouter key。留空則略過。 - key: DEEPSEEK_API_KEY type: STRING name: DeepSeek API Key（選填） description: 選填。上游 DeepSeek key。留空則略過。 readme: |- # Bifrost [Bifrost](https://github.com/maximhq/bifrost) 是開源 AI 閘道。本模板依官方 [runtime contract](https://docs.getbifrost.ai/deployment-guides/runtime-contract)（映像 `v2.0.0`）部署：單一 Go 行程、SQLite 落在 `/app/data`，控制台與推論共用同一個 HTTP 連接埠。 包含服務： - **bifrost** — `maximhq/bifrost:v2.0.0`（API、控制台、SQLite 設定庫、SQLite 日誌庫） 不需要 PostgreSQL、Redis 或額外 worker。OSS 的 SQLite 只支援單一副本。 ## 資源建議 映像約 87 MB。個人或小團隊 0.5–1 vCPU、512 MB–1 GB RAM 即可。 ## 設定 需要填寫閘道網域。控制台驗證開箱即開（使用者名稱 `admin`，密碼自動產生）。到 bifrost 服務的 env 查看 `BIFROST_ADMIN_PASSWORD` 與 `BIFROST_VIRTUAL_KEY`。 可選上游 key（`OPENAI_API_KEY`、`ANTHROPIC_API_KEY`、`OPENROUTER_API_KEY`、`DEEPSEEK_API_KEY`）透過 `config.json` 的 `env.*` 引用注入。留空也可以，之後在控制台補。Gemini、Azure、Bedrock、Groq、Ollama 等可在 UI 新增，不必重新部署。 `BIFROST_ENCRYPTION_KEY` 會自動產生，重啟後不要改。改了會導致已加密的設定讀不出來。 ## 首次登入 打開閘道網域，使用： - 使用者名稱：`admin` - 密碼：env 裡的 `BIFROST_ADMIN_PASSWORD` 然後可在控制台繼續加虛擬 Key、預算或其他供應商。 ## 快速測試 `/health` 回傳 200，且至少設定了一個上游 key 之後： ```bash curl --fail-with-body --max-time 60 \ -X POST https:///v1/chat/completions \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{"model":"openai/gpt-4o-mini","messages":[{"role":"user","content":"Hello"}]}' ``` 管理員 Basic Auth 也可以：`-u admin:`。 對接 OpenAI SDK / GapSeek 的 `OPENAI_BASE_URL`： ```text https:///openai ``` `OPENAI_API_KEY` 填 `BIFROST_VIRTUAL_KEY`。模型名稱是 `provider/model`，例如 `openai/gpt-4o-mini`、`anthropic/claude-sonnet-4-5`、`openrouter/deepseek/deepseek-chat`。 ## 本模板範圍 | 需求 | 本模板 | | --- | --- | | OpenAI 相容 `/v1` + `/openai` | 已包含 | | 內建控制台 | 已包含（同一網域） | | 故障轉移 / 負載平衡 / 虛擬 Key | 已包含（在 UI 設定） | | 語意快取、MCP、更多供應商 | 在控制台新增 | | 多副本 / HA | 不包含；需要 Bifrost Enterprise + PostgreSQL | 不要把未驗證的閘道暴露到公網。推論請求預設需要虛擬 Key（或管理員 Basic Auth），除非你在 Security 關掉 inference 驗證。 ## 版本 映像以 digest 鎖定 `maximhq/bifrost:v2.0.0`（拉取日期 2026-09-03）。 ## 參考 - 安裝: https://docs.getbifrost.ai/quickstart/gateway/setting-up - Runtime contract: https://docs.getbifrost.ai/deployment-guides/runtime-contract - config.json: https://docs.getbifrost.ai/deployment-guides/config-json - GitHub: https://github.com/maximhq/bifrost - License: Apache-2.0