Clonvo: the AI gateway for agents
The AI gateway for agents: 150+ models plus web search, scraping and research tools on one API key. Pay at provider list prices or bring your own keys.
One API key for 150+ models and 22 tools your agents call to search, browse, run code and look up companies.
Get an API key or read the docs.
Quick start
Use the OpenAI SDK with Clonvo’s base URL. To switch models, change the model id.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.clonvo.com/v1",
api_key=os.environ["CLONVO_API_KEY"],
)
completion = client.chat.completions.create(
model="openai/gpt-6.1-sol",
messages=[{"role": "user", "content": "Summarise the plot of Hamlet in one sentence."}],
)
print(completion.choices[0].message.content)
Example model ids:
openai/gpt-6.1-sol(GPT-6.1 Sol)anthropic/claude-sonnet-5-5(Claude Sonnet 5.5)google/gemini-3.8-flash(Gemini 3.8 Flash)x-ai/grok-4-7(Grok 4.7)deepseek/deepseek-v4-pro(DeepSeek V4 Pro)mistralai/mistral-large-4(Mistral Large 4)moonshotai/kimi-k3.1-pro(Kimi K3.1 Pro)meta/muse-spark-1.3(Muse Spark 1.3)qwen/qwen3.8-max(Qwen3.8 Max)
Models and tools on one key
Models and tools share one API key, one credit balance and one activity log. Adding a capability means changing a string, not signing up for another provider.
- 150+ models (
POST /v1/chat/completions): Chat Completions for OpenAI, Anthropic, Google and 8 more providers. Switch models by changing the model id. Docs. - Web tools (
POST /v1/tools/web_search): Search, read and crawl the web, extract typed JSON, and get cited answers and research reports. Docs. - Companies and people (
POST /v1/tools/company_enrich): Company profiles, news, jobs and lookalikes, and professional profiles by role or company. Docs. - MCP server (
https://api.clonvo.com/mcp): Sign in from ChatGPT, Claude, Cursor or any MCP client, and its agent gets every Clonvo tool. Docs.
11 providers, plus free models
- OpenAI: 26 models
- Anthropic: 17 models
- Google Gemini: 14 models
- xAI Grok: 7 models
- DeepSeek: 2 models
- Mistral: 7 models
- Moonshot Kimi: 3 models
- Meta: 5 models
- Alibaba Qwen: 19 models
- TypeSafe: 2 models
- Parallel: 1 model
Browse the model catalogue, with prices and context lengths for each model.
What you get
Routing, keys and spend controls
Keep your OpenAI code
Tool calling, streaming, structured outputs and reasoning use one request shape for every provider, Anthropic and Gemini included.
Fall back to another model
Pass models: [...] and Clonvo tries them in order when a model is rate limited or a provider returns a server error.
API keys per project, with limits
Group keys into projects, cap each one daily, weekly or monthly, and revoke one without touching the rest.
Built for teams
Invite owners, admins and members, and organise them into teams. Credits, keys and usage belong to the organization.
See where the money goes
Every request is logged with its model, provider, tokens and cost, and spend breaks down by project, key and model.
Prompts aren’t stored by default
Clonvo keeps request metadata, not prompts or responses, unless your organization turns on retention.
Questions
Do I need to change my code?
Point the OpenAI SDK at https://api.clonvo.com/v1, use a Clonvo API key, and set model to any id from the catalogue, such as anthropic/claude-sonnet-5-5. The request and response shapes are OpenAI’s Chat Completions format.
What happens when a provider returns an error?
Add fallback models with models: [...] and Clonvo tries them in order when a request is rate limited or fails with a server error. If your own key fails, Clonvo retries with credits unless you’ve set that key to “Always use this key”. Invalid requests (400s) are returned straight away. Read about fallbacks.
Do you store my prompts?
Not by default. Clonvo records metadata for each request (model, tokens, cost, timing), not the prompt or response. An organization admin can turn on retention to keep request and response bodies for their own Activity view. Free models are the exception: once an owner or admin accepts their terms, Clonvo stores free-model prompts and responses to train and evaluate AI models. Data and privacy explains both settings.
Are there free models?
Yes. Model ids ending in :free are served through OpenRouter at no cost, limited to 25 requests a minute and 1,000 a day per organization on Free, with no limit once it has bought credits. To use them, an owner or admin turns them on in your organization’s privacy settings and accepts their terms. Free-model prompts and responses are stored by Clonvo for up to 12 months and may be used, or licensed with personal data removed, to train AI models. Paid models with a data-sharing version (Meta’s Muse Spark) run on it at a lower price, and Meta may train on those requests; Clonvo doesn’t store them.
More
- Pricing: plans, fees and bring-your-own-key costs.
- Quickstart: make your first request.