Dhenara is an open-source Python package designed for seamless interaction with a variety of AI models through a unified API. This lightweight framework simplifies the integration of multiple AI models into Python applications, prioritizing simplicity and minimal dependencies, while ensuring type safety through the use of Pydantic Models.

Key Features

• Genuinely Open Source: Developed as a community resource with no internal tool limitations.
• Unified API: Engage with numerous AI providers through a consistent interface for streamlined integration.
• Type Safety: Leverage Pydantic for robust type checks and validation to enhance code reliability.
• Effortless Model Regeneration: With a unified Pydantic output and built-in prompt formatting, effortlessly transition output from one model to another.
• Streaming Support: Enjoy first-class support for streaming responses, including accumulated responses to match non-streaming outputs.
• Asynchronous Functionality: Supports both synchronous and asynchronous operations for optimal flexibility in various applications.
• Resource Management: Automatically manage connections, retries, and timeouts to enhance reliability and performance.
• Pre-Configured Models: Access an array of foundation models with sensible default settings for quicker startup.
• Test Mode: Enable test responses for both streaming and non-streaming generation, streamlining the testing process.
• Cost and Usage Insights: Obtain derived cost and usage data with optional model endpoint charges for commercial deployment.
• Community-Centric Design: Features a modular architecture that separates API credentials, models, and configurations to facilitate flexible scaling and deployment options.

Example Usage

Here’s a straightforward example demonstrating how to utilize Dhenara to interact with an AI model. More examples can be found in the extensive documentation at docs.dhenara.com.

from dhenara.ai import AIModelClient
from dhenara.ai.types import AIModelCallConfig, AIModelEndpoint
from dhenara.ai.types.external_api import AIModelAPIProviderEnum
from dhenara.ai.types.genai import AIModelAPI
from dhenara.ai.types.genai.foundation_models.anthropic.chat import Claude37Sonnet

# Create an API
api = AIModelAPI(
    provider=AIModelAPIProviderEnum.ANTHROPIC,
    api_key="your_api_key",
)

# Create an endpoint using a pre-configured model
model_endpoint = AIModelEndpoint(
    api=api,
    ai_model=Claude37Sonnet,
)

# Configure the API call
config = AIModelCallConfig(
    max_output_tokens=16000,
    reasoning=True,  # Thinking/reasoning mode
    max_reasoning_tokens=8000,
    streaming=False,
)

# Create the client
client = AIModelClient(
    model_endpoint=model_endpoint,
    config=config,
    is_async=False,
)

# Create a prompt
prompt = {
    "role": "user",
    "content": "Explain quantum computing in simple terms",
}

# Generate a response
response = client.generate(prompt=prompt)

# If not streaming
if response.chat_response:
    print(response.chat_response.choices[0].contents[0].get_text())

# If streaming
elif response.stream_generator:
    for chunk, _ in response.stream_generator:
        if chunk:
            print(
                chunk.data.choice_deltas[0].content_deltas[0].get_text_delta(),
                end="",
                flush=True,
            )

For comprehensive documentation, visit docs.dhenara.com. This resource provides all the information necessary to maximize the potential of Dhenara in Python applications.