Environment:CrewAIInc CrewAI LLM Provider Credentials

Knowledge Sources	CrewAI CrewAI Docs
Domains	Infrastructure, LLM_Integration
Last Updated	2026-02-11 17:00 GMT

Overview

API credentials environment for 12+ LLM providers supported by CrewAI, including OpenAI, Anthropic, Google Gemini, AWS Bedrock, and Azure.

Description

CrewAI supports multiple LLM providers through environment variables. At minimum, one provider's API key must be configured. The default provider is OpenAI, but the framework also has native providers for Anthropic, Google Gemini, AWS Bedrock, and Azure AI Inference. Additional providers are available through the LiteLLM integration layer. Each provider requires specific environment variables for authentication.

Usage

Use this environment for any CrewAI workflow that involves LLM calls. This covers virtually all Agent, Crew, and Flow operations. The specific credentials needed depend on which `model` string is configured on your Agents or Crew.

System Requirements

Category	Requirement	Notes
Network	Internet access	Required for all cloud LLM API calls
Provider Account	At least one LLM provider account	OpenAI is the default provider

Dependencies

Core (included in base install)

`openai` >=1.83.0,<3 (default provider)

Optional Provider Packages

`anthropic` ~=0.73.0 (install via `crewai[anthropic]`)
`google-genai` ~=1.49.0 (install via `crewai[google-genai]`)
`boto3` ~=1.40.45 (install via `crewai[bedrock]`)
`azure-ai-inference` ~=1.0.0b9 (install via `crewai[azure-ai-inference]`)
`litellm` >=1.74.9,<3 (install via `crewai[litellm]`)
`ibm-watsonx-ai` ~=1.3.39 (install via `crewai[watson]`)

Credentials

OpenAI (Default Provider)

`OPENAI_API_KEY`: OpenAI API key (required if using OpenAI models)
`OPENAI_BASE_URL`: Custom base URL for OpenAI-compatible APIs
`OPENAI_MODEL_NAME`: Override default model name

Anthropic

`ANTHROPIC_API_KEY`: Anthropic API key (required for Claude models)

Google Gemini

`GOOGLE_API_KEY` or `GEMINI_API_KEY`: Google API key
`GOOGLE_CLOUD_PROJECT`: GCP project ID (for Vertex AI)
`GOOGLE_CLOUD_LOCATION`: GCP region (defaults to `"us-central1"`)
`GOOGLE_GENAI_USE_VERTEXAI`: Set to `"true"` to use Vertex AI instead of AI Studio

AWS Bedrock

`AWS_ACCESS_KEY_ID`: AWS access key
`AWS_SECRET_ACCESS_KEY`: AWS secret key
`AWS_SESSION_TOKEN`: AWS session token (optional, for temporary credentials)
`AWS_REGION_NAME`: AWS region

Azure

`AZURE_API_KEY`: Azure API key
`AZURE_ENDPOINT` or `AZURE_OPENAI_ENDPOINT` or `AZURE_API_BASE`: Azure endpoint URL
`AZURE_API_VERSION`: API version (defaults to `"2024-06-01"`)

IBM Watson

`WATSONX_URL`: WatsonX service URL
`WATSONX_APIKEY`: WatsonX API key
`WATSONX_PROJECT_ID`: WatsonX project ID

Other Providers

`NVIDIA_NIM_API_KEY`: NVIDIA NIM API key
`GROQ_API_KEY`: Groq API key
`HF_TOKEN`: Hugging Face token
`SAMBANOVA_API_KEY`: SambaNova API key
`CEREBRAS_API_KEY`: Cerebras API key

Model Configuration

`MODEL` or `MODEL_NAME`: Override default model name
`BASE_URL` or `API_BASE`: Override default API base URL

Quick Install

# Default (OpenAI only)
uv add crewai

# With Anthropic native provider
uv add "crewai[anthropic]"

# With Google Gemini native provider
uv add "crewai[google-genai]"

# With AWS Bedrock native provider
uv add "crewai[bedrock]"

# With Azure AI Inference native provider
uv add "crewai[azure-ai-inference]"

# With LiteLLM for 100+ providers
uv add "crewai[litellm]"

Code Evidence

OpenAI API key requirement from `lib/crewai/src/crewai/llms/providers/openai/completion.py:336-338`:

self.api_key = os.getenv("OPENAI_API_KEY")
if self.api_key is None:
    raise ValueError("OPENAI_API_KEY is required")

Anthropic API key requirement from `lib/crewai/src/crewai/llms/providers/anthropic/completion.py:225-227`:

self.api_key = os.getenv("ANTHROPIC_API_KEY")
if self.api_key is None:
    raise ValueError("ANTHROPIC_API_KEY is required")

Gemini multi-key support from `lib/crewai/src/crewai/llms/providers/gemini/completion.py:113-119`:

api_key or os.getenv("GOOGLE_API_KEY") or os.getenv("GEMINI_API_KEY")
self.project = project or os.getenv("GOOGLE_CLOUD_PROJECT")
self.location = location or os.getenv("GOOGLE_CLOUD_LOCATION") or "us-central1"
use_vertexai = os.getenv("GOOGLE_GENAI_USE_VERTEXAI", "").lower() == "true"

Optional provider import pattern from `lib/crewai/src/crewai/llms/providers/anthropic/completion.py:23-31`:

try:
    from anthropic import Anthropic, AsyncAnthropic, transform_schema
except ImportError:
    raise ImportError(
        'Anthropic native provider not available, to install: uv add "crewai[anthropic]"'
    ) from None

CLI provider enumeration from `lib/crewai/src/crewai/cli/constants.py:10-115`:

ENV_VARS: dict[str, list[dict[str, Any]]] = {
    "openai": [{"prompt": "Enter your OPENAI API key", "key_name": "OPENAI_API_KEY"}],
    "anthropic": [{"prompt": "Enter your ANTHROPIC API key", "key_name": "ANTHROPIC_API_KEY"}],
    "gemini": [{"prompt": "Enter your GEMINI API key", "key_name": "GEMINI_API_KEY"}],
    # ... 12+ providers total
}

Common Errors

Error Message	Cause	Solution
`ValueError: OPENAI_API_KEY is required`	No OpenAI key set	Set `OPENAI_API_KEY` in `.env` or environment
`ValueError: ANTHROPIC_API_KEY is required`	No Anthropic key set	Set `ANTHROPIC_API_KEY` in `.env` or environment
`ImportError: Anthropic native provider not available`	Missing optional dependency	Run `uv add "crewai[anthropic]"`
`ImportError: AWS Bedrock native provider not available`	Missing boto3	Run `uv add "crewai[bedrock]"`
`ImportError: Azure AI Inference native provider not available`	Missing azure-ai-inference	Run `uv add "crewai[azure-ai-inference]"`
`ImportError: Google Generative AI provider not available`	Missing google-genai	Run `uv add "crewai[google-genai]"`

Compatibility Notes

Default Provider: OpenAI is the default; its SDK is included in core dependencies
Native Providers: Anthropic, Gemini, Bedrock, and Azure have native integrations with direct SDK calls
LiteLLM Fallback: For providers without native support, install `crewai[litellm]` to access 100+ providers
Ollama (Local): No API key needed; defaults to `http://localhost:11434` as base URL
Environment Files: CrewAI loads `.env` files via `python-dotenv` automatically

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment