For AI agent (llms.txt) | 中文

ai-sdk-provider-env

Environment-variable-driven provider for Vercel AI SDK. Switch AI providers and models without code changes.

Quick Start

pnpm add ai-sdk-provider-env

# .env
OPENAI_API_KEY=sk-xxx

import { generateText } from 'ai'
import { envProvider } from 'ai-sdk-provider-env'

const provider = envProvider()

const { text } = await generateText({
  model: provider.languageModel('openai/gpt-4o'),
  prompt: 'Hello!',
})

The config set name openai auto-matches the built-in preset — only an API key is needed.

Any valid env var prefix becomes a config set. Two endpoints, zero code changes:

# .env
FAST_BASE_URL=https://fast-api.example.com/v1
FAST_API_KEY=key-fast

SMART_BASE_URL=https://smart-api.example.com/v1
SMART_API_KEY=key-smart

provider.languageModel('fast/llama-3-8b')
provider.languageModel('smart/gpt-4o')

Environment Variables

Model ID format: {configSet}/{modelId}. The config set maps to an uppercased env var prefix.

Config set names must match [A-Za-z_][A-Za-z0-9_-]* — ASCII letters, digits, underscores, and hyphens. Hyphens are automatically normalized to underscores for env var lookup:

# Config set "my-api" → reads MY_API_* env vars
MY_API_BASE_URL=https://api.example.com/v1
MY_API_API_KEY=sk-xxx

provider.languageModel('my-api/some-model')  // reads MY_API_* env vars
provider.languageModel('my_api/some-model')  // same env vars

For config set names outside these rules (e.g. Unicode, dots), use the configs option instead.

| Variable | Required | Description | |---|---|---| | {PREFIX}_API_KEY | Yes | API key | | {PREFIX}_BASE_URL | Unless preset matches | API base URL | | {PREFIX}_PRESET | No | Built-in preset name (e.g. openai) | | {PREFIX}_COMPATIBLE | No | openai · anthropic · gemini · openai-compatible (default) | | {PREFIX}_HEADERS | No | Custom HTTP headers (JSON) | | {PREFIX}_NATIVE_ROUTING | No | Enable/disable native model routing (true/false) |

When _PRESET is set or auto-detected, _BASE_URL and _COMPATIBLE fall back to preset defaults.

Compatibility modes

| Value | SDK | Fallback | |---|---|---| | openai | @ai-sdk/openai | @ai-sdk/openai-compatible if not installed | | anthropic | @ai-sdk/anthropic | None | | gemini | @ai-sdk/google | None | | openai-compatible | @ai-sdk/openai-compatible (default) | — |

Install provider SDKs as needed: pnpm add @ai-sdk/openai @ai-sdk/anthropic @ai-sdk/google

Built-in Presets

When the config set name matches a preset, it auto-applies — only _API_KEY is needed:

DEEPSEEK_API_KEY=sk-xxx          # config set "deepseek" matches the preset

provider.languageModel('deepseek/deepseek-chat')   // just works

| Preset | Base URL | Compatible | |---|---|---| | openai | https://api.openai.com/v1 | openai | | anthropic | https://api.anthropic.com | anthropic | | google | https://generativelanguage.googleapis.com/v1beta | gemini | | opencode-zen | https://opencode.ai/zen/v1 | openai-compatible (nativeRouting enabled) | | opencode-go | https://opencode.ai/zen/go/v1 | openai-compatible | | deepseek | https://api.deepseek.com | openai-compatible | | groq | https://api.groq.com/openai/v1 | openai-compatible | | together | https://api.together.xyz/v1 | openai-compatible | | fireworks | https://api.fireworks.ai/inference/v1 | openai-compatible | | mistral | https://api.mistral.ai/v1 | openai-compatible | | moonshot | https://api.moonshot.ai/v1 | openai-compatible | | moonshot-china | https://api.moonshot.cn/v1 | openai-compatible | | perplexity | https://api.perplexity.ai | openai-compatible | | openrouter | https://openrouter.ai/api/v1 | openai-compatible | | siliconflow | https://api.siliconflow.com/v1 | openai-compatible | | siliconflow-china | https://api.siliconflow.cn/v1 | openai-compatible | | xai | https://api.x.ai/v1 | openai-compatible | | zai | https://api.z.ai/api/paas/v4 | openai-compatible | | zhipu | https://open.bigmodel.cn/api/paas/v4 | openai-compatible |

To disable auto-detection: envProvider({ presetAutoDetect: false }). See Advanced Usage for details.

Native Routing

When a config set uses a gateway that exposes multiple AI providers (like opencode-zen), nativeRouting auto-detects the model family from the model ID prefix and routes it to the appropriate native SDK:

• claude-* → @ai-sdk/anthropic
• gemini-* → @ai-sdk/google
• gpt-* → @ai-sdk/openai
• Other models fall back to the config set's default compatible mode

# opencode-zen preset has nativeRouting enabled — just set API key
OPENCODE_ZEN_API_KEY=zen-xxx

// Routes to native Anthropic SDK automatically
provider.languageModel('opencode-zen/claude-sonnet-4-20250514')

// Routes to native Google SDK automatically
provider.languageModel('opencode-zen/gemini-3-flash')

// Other models use openai-compatible
provider.languageModel('opencode-zen/minimax-m2.5')

To disable: OPENCODE_ZEN_NATIVE_ROUTING=false

Known limitation: o1-*, o3-*, chatgpt-* models are not automatically routed. Use {PREFIX}_COMPATIBLE=openai explicitly for these.

Documentation

• API Reference — envProvider() options, types, model ID format
• Advanced Usage — Code-based configs, custom fetch/headers, custom separator, provider registry
• Bundler Usage — For bun build, vite build, and other bundlers