The only AI APIyour agents need
New models the day they ship. Chat, image, audio, and video behind one key, with zero data retention.
- Free tier, no card
- OpenAI-compatible
- Zero data retention
- 4M+ users
Frontier and open-source, same endpoint
Capabilities
All under one API key
Text, code, images, video, speech, music, embeddings, and live web search, behind one private endpoint. Same auth, same shapes, same billing.
Quickstart
Only the base URL changes
Point the OpenAI SDK at Venice and your app just works. Streaming, function calling, and structured outputs keep the same shape.
One line. That's the whole migration.
from openai import OpenAI
client = OpenAI(
api_key=os.environ["VENICE_API_KEY"],
base_url="https://api.venice.ai/api/v1",
)
response = client.chat.completions.create(
model="venice-uncensored",
messages=[{"role": "user", "content": "Hello, Venice."}],
extra_body={
"venice_parameters": {
"enable_web_search": "auto",
"enable_web_citations": True,
}
},
)
print(response.choices[0].message.content)Agent ready
Bring your own agent
Point the agents you already use at Venice. Native providers or the OpenAI-compatible API. Every request private and uncensored.
setup-mcp venice.ai/mcpIt reads the docs and wires itself up
First-party and open source
Wire Venice into your stack.
Terminal, MCP client, or Agent Skills. Take your pick.
Privacy
Zero retention by default, provable on demand
Venice never stores your prompts or responses. Add TEE or E2EE when you need privacy you can verify.
We never store your prompts
Requests are processed in memory and discarded. Nothing is logged, trained on, or sold.
Providers see tokens, not you
Frontier models are reached through Venice, so upstream providers never receive your identity.
Run inside a hardware enclave
TEE inference attests to the code handling your request. Verify the signature yourself.
Encrypt before it leaves your app
With end-to-end encryption, only the client holds the key. Venice cannot read the request.
Ask without guardrails
Build uncensored
No filters, no rewrites, no moralizing. Unrestricted access to models that just answer, private by default.
Pay per token / Pro from $18 / Stake for free access