XavaInferencesign indocs
agents · plugin · x402

Give your agent every model, pay per call.

One install puts 229 text, image, video and audio models inside your coding agent, each cheaper than going direct. Allocate your discount as savings or token buybacks. Chat runs on your key; images, video and audio can be paid per call in USDC, with no account at all.

229models, one MCP server
10AI clients, one command each
117media models payable per call
0accounts needed for x402
select your agent

One command. Every model.

Pick your client and paste the line. The plugin source is public at github.com/xavalabs/xinf-plugin.

Paste this in your terminal.

terminal
claude plugin marketplace add xavalabs/xinf-plugin && claude plugin install xinf@xinf

Then set your key: export XINF_API_KEY=xk_live_... (skip it to use only the catalog and x402 media). This deployment is not the default origin, so also: export XINF_BASE_URL=https://zinf.dev

More setup for Claude Code →

x402

Pay per call. No account.

Your agent can pay for each image, video or audio clip by itself, in USDC, from its own wallet. No sign-up and no top-up.

installlive
01

Add our plugin to your agent with one copy and paste.

give it a wallet
02

Point it at a Solana wallet that holds a little USDC. Keep this wallet just for your agent.

set a limit
03

Tell it the most it may spend per call and per session. It asks you before going over.

terminal, then your shell profile
claude plugin marketplace add xavalabs/xinf-plugin && claude plugin install xinf@xinf

export XINF_X402_KEYPAIR=~/.config/solana/xinf-agent.json
export XINF_X402_MAX_PER_CALL_USD=0.25
export XINF_X402_SESSION_BUDGET_USD=2

Restart Claude Code and ask it for an image. It shows the price and pays within your limit.

Works for 117 image, video and audio models. How it works under the hood: x402 docs.

agentic apps

Build agentic apps with one key.

Use the same key in your own apps and agents instead of a chat window. Every model, cheaper than going direct, through any OpenAI-compatible SDK.

get a key

Create an account and an API key, then set it where your app runs.

in your code
python
from openai import OpenAI

client = OpenAI(
    base_url="https://zinf.dev/v1",
    api_key="xk_live_...",  # your Xava Inference key
)

r = client.chat.completions.create(
    model="openai/gpt-5.4",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(r.choices[0].message.content)
faq

Questions from agent builders.

What does the plugin do?
It lets your AI agent use every model we offer: it can write, make images, video and audio, check prices and see your balance, all from inside the agent you already use. Details are in the agent docs.
Do I need an account?
Not for images, video and audio: your agent can pay for each one from its own wallet with x402. For chat and text, create a free account and an API key, then set XINF_API_KEY.
What is x402?
A way for software to pay for things on the web by itself. Your agent asks for an image, sees the exact price, pays it in USDC, and gets the result. No sign-up, no subscription, no leftover balance.
Can my agent overspend?
No. Every payment is for one exact price, and you set a spending limit. Your agent asks you before going over it. Keep only a little USDC in its wallet.
Which agents are supported?
Claude Code, Codex, Cursor, Gemini CLI, OpenCode, Kimi, Pi, Cline, Windsurf, Any OpenAI SDK. Any other MCP client works too; see the MCP docs.
Where does my API key live?
In the XINF_API_KEY environment variable on your computer, never in a file you share. Cline is the one exception: paste the key into its MCP settings.
Do you store prompts or outputs?
No prompts, ever. We keep billing details only: model, amount and time. Generated media is stored so its link keeps working, and the link is private.
docs

Everything else is in the docs.

Per-client setup, the MCP tool reference and the full x402 exchange.