GitHub - actuallyrizzn/Venice-AI-SDK: Comprehensive Python SDK for Venice AI - Advanced LLM, Image Generation, Audio Synthesis, and AI Assistant APIs with Full Swagger Specification Support

from venice_sdk import VeniceClient, create_client

# Initialize the client (loads from VENICE_API_KEY env var)
client = VeniceClient()

# Or create with explicit API key
# client = create_client(api_key="your-api-key")

# Chat completions
response = client.chat.complete(
    messages=[
        {"role": "user", "content": "Hello! What can you help me with?"}
    ],
    model="llama-3.3-70b",
    temperature=0.7,
    max_completion_tokens=500
)
print(response.choices[0].message.content)

# Image generation
image = client.images.generate(
    prompt="A serene mountain landscape at sunset",
    model="dall-e-3"
)
image.save("mountain.png")

# Text-to-speech
audio = client.audio.speech(
    input_text="Hello from Venice AI!",
    voice="alloy"
)
audio.save("hello.mp3")

# Generate embeddings
embeddings = client.embeddings.generate([
    "The quick brown fox jumps over the lazy dog",
    "A fast brown fox leaps over a sleepy dog"
])
similarity = embeddings[0].cosine_similarity(embeddings[1])
print(f"Similarity: {similarity}")

# Video generation
job = client.video.queue(
    model="kling-2.6-pro-text-to-video",
    prompt="A serene sunset over a calm ocean",
    duration=5,
    aspect_ratio="16:9"
)
completed_job = client.video.wait_for_completion(job.job_id)
if completed_job.is_completed():
    completed_job.download("video.mp4")

# List available characters
characters = client.characters.list()
for char in characters[:3]:
    print(f"- {char.name}: {char.description}")

# Check account usage
usage = client.billing.get_usage()
print(f"Credits remaining: {usage.credits_remaining}")

# Advanced usage with pagination
from datetime import datetime, timedelta

end_date = datetime.now()
start_date = end_date - timedelta(days=30)

usage = client.billing.get_usage(
    currency="USD",
    start_date=start_date,
    end_date=end_date,
    limit=50,
    page=1,
    sort_order="desc"
)

# Access pagination info
if usage.pagination:
    print(f"Page {usage.pagination['page']} of {usage.pagination['totalPages']}")
    print(f"Total records: {usage.pagination['total']}")

# Generate an image
image = client.images.generate(
    prompt="A futuristic cityscape at night",
    model="dall-e-3",
    size="1024x1024",
    quality="hd"
)

# Edit an existing image
edited = client.image_edit.edit(
    image="path/to/image.png",
    prompt="Add a rainbow in the sky",
    model="dall-e-2-edit"
)

# Upscale an image
upscaled = client.image_upscale.upscale(
    image="path/to/small_image.png",
    scale=2
)

# List available styles
styles = client.image_styles.list_styles()
for style in styles:
    print(f"{style.name}: {style.description}")

# Convert text to speech
audio = client.audio.speech(
    input_text="Hello, this is a test of the Venice AI text-to-speech system.",
    voice="alloy",
    response_format="mp3",
    speed=1.0
)

# Save to file
audio.save("speech.mp3")

# List available voices
voices = client.audio.get_voices()
for voice in voices:
    print(f"{voice.name}: {voice.description}")

# Batch processing
from venice_sdk import AudioBatchProcessor
processor = AudioBatchProcessor(client.audio)
texts = ["Hello", "World", "Venice AI"]
saved_files = processor.process_batch(texts, "output_dir/")

# List all characters
characters = client.characters.list()

# Search for specific characters
assistants = client.characters.search("assistant")

# Get a specific character
venice_char = client.characters.get("venice")

# Use character in chat
response = client.chat.complete(
    messages=[{"role": "user", "content": "Tell me about yourself"}],
    model="llama-3.3-70b",
    venice_parameters={"character_slug": "venice"}
)

# Generate embeddings
embeddings = client.embeddings.generate([
    "Machine learning is fascinating",
    "AI will change the world",
    "The weather is nice today"
])

# Calculate similarity
similarity = embeddings[0].cosine_similarity(embeddings[1])
print(f"Similarity: {similarity}")

# Semantic search
from venice_sdk import SemanticSearch
search = SemanticSearch(client.embeddings)
search.add_documents([
    "Python is a programming language",
    "Machine learning uses algorithms",
    "The sky is blue"
])

results = search.search("programming", top_k=2)
for result in results:
    print(f"{result['similarity']:.3f}: {result['document']}")

# Text-to-video generation
job = client.video.queue(
    model="kling-2.6-pro-text-to-video",
    prompt="A cat playing with a ball of yarn in slow motion",
    duration=5,
    aspect_ratio="16:9",
    resolution="1080p"
)

# Wait for completion with progress tracking
def track_progress(job):
    print(f"Status: {job.status} - {job.progress}%")

completed_job = client.video.wait_for_completion(
    job.job_id,
    poll_interval=5,
    callback=track_progress
)

if completed_job.is_completed():
    completed_job.download("cat_video.mp4")

# Get price quote before generating
quote = client.video.quote(
    model="kling-2.6-pro-text-to-video",
    prompt="A beautiful sunset",
    duration=10,
    resolution="1080p"
)
print(f"Estimated cost: ${quote.estimated_cost} {quote.currency}")

# Image-to-video animation
image = client.images.generate(
    prompt="A serene mountain landscape",
    model="dall-e-3"
)

animated_job = client.video.queue(
    model="kling-2.6-pro-image-to-video",
    image=image.url,
    prompt="Animate with gentle movement",
    duration=5
)

completed_animated = client.video.wait_for_completion(animated_job.job_id)
if completed_animated.is_completed():
    completed_animated.download("animated_landscape.mp4")

# Get account usage
usage = client.billing.get_usage()
print(f"Total usage: {usage.total_usage}")
print(f"Credits remaining: {usage.credits_remaining}")

# Check rate limits
limits = client.api_keys.get_rate_limits()
print(f"Requests per minute: {limits.requests_per_minute}")

# Generate Web3 API key
web3_key = client.api_keys.generate_web3_key(
    name="Web3 Integration",
    description="For blockchain applications"
)
print(f"Web3 Key: {web3_key.api_key}")

# Get account summary
summary = client.get_account_summary()
print(f"Account Summary: {summary}")

# List all API keys
api_keys = client.api_keys.list()
for key in api_keys:
    print(f"Key: {key.name} ({key.id})")
    print(f"  Type: {key.permissions.get('type', 'Unknown')}")
    print(f"  Created: {key.created_at}")

# Create a new API key
new_key = client.api_keys.create(
    name="My New Key",
    permissions=["read", "write"]
)
print(f"Created key: {new_key.id}")

# Get specific API key details
key_details = client.api_keys.get(new_key.id)
print(f"Key details: {key_details}")

# Delete an API key
success = client.api_keys.delete(new_key.id)
print(f"Key deleted: {success}")

# Get comprehensive billing information
billing_summary = client.billing.get_billing_summary()
print(f"Billing Summary: {billing_summary}")

# Get detailed usage by model
model_usage = client.billing.get_usage_by_model()
for model, usage in model_usage.items():
    print(f"{model}: {usage.requests} requests, {usage.tokens} tokens")

# Get model traits
traits = client.models_traits.get_traits()
for model_id, model_traits in traits.items():
    print(f"{model_id}:")
    print(f"  Function calling: {model_traits.supports_function_calling()}")
    print(f"  Web search: {model_traits.supports_web_search()}")

# Get compatibility mapping
mapping = client.models_compatibility.get_mapping()
venice_model = mapping.get_venice_model("gpt-3.5-turbo")
print(f"OpenAI gpt-3.5-turbo maps to: {venice_model}")

# Find models by capability
function_models = client.models_traits.find_models_by_capability("function_calling")
print(f"Models with function calling: {function_models}")

# Set your API key
venice auth your-api-key-here

# Check authentication status
venice status

# Advanced parameter control
response = client.chat.complete(
    messages=[{"role": "user", "content": "Write a creative story"}],
    model="llama-3.3-70b",
    temperature=1.2,
    frequency_penalty=0.3,
    presence_penalty=-0.1,
    repetition_penalty=1.1,
    max_completion_tokens=500,
    n=3,  # Generate 3 different responses
    seed=42,  # For reproducible results
    stop=["END", "STOP"]  # Stop sequences
)

# Dynamic temperature scaling
response = client.chat.complete(
    messages=[{"role": "user", "content": "Write a poem"}],
    model="llama-3.3-70b",
    min_temp=0.5,
    max_temp=1.5,
    logprobs=True,
    top_logprobs=3
)

# Stream chat responses in real-time
for chunk in client.chat.complete_stream(
    messages=[{"role": "user", "content": "Tell me a story about a robot"}],
    model="llama-3.3-70b"
):
    print(chunk, end="", flush=True)

# Or use the streaming method directly
response = client.chat.complete(
    messages=[{"role": "user", "content": "Hello!"}],
    model="llama-3.3-70b",
    stream=True,
    stream_options={"include_usage": True}
)

for chunk in response:
    if chunk.get("choices") and chunk["choices"][0].get("delta", {}).get("content"):
        print(chunk["choices"][0]["delta"]["content"], end="", flush=True)

# Define tools
tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get the current weather in a location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string",
                        "description": "The city and state, e.g. San Francisco, CA"
                    }
                },
                "required": ["location"]
            }
        }
    }
]

# Use tools in chat
response = chat.complete(
    messages=[{"role": "user", "content": "What's the weather in San Francisco?"}],
    model="llama-3.3-70b",
    tools=tools
)

from venice_sdk.errors import VeniceAPIError, RateLimitError, UnauthorizedError

try:
    response = chat.complete(...)
except RateLimitError:
    print("Rate limit exceeded. Please try again later.")
except UnauthorizedError:
    print("Authentication failed. Please check your API key.")
except VeniceAPIError as e:
    print(f"API error: {e}")

pip install -e ".[docs]"
mkdocs serve