WRAITH Nexus

Every AI model. One key. Paid on Monad.

An OpenAI-compatible API for 250+ models: Claude, Gemini, Grok, GLM, Qwen, DeepSeek, FLUX, Whisper and more, from private Venice inference and NEAR AI confidential compute. Deposit USDC on Monad, register a key from your wallet, and go. Agents can skip the key and pay per call with x402.

Get an API key Setup in 2 minutes For agents (x402)

250+models, five modalities

USDCon Monad, prepaid, no card

x402pay-per-call for agents

On chainevery call recorded and anchored

Drop-in

Change two lines. Keep your code.

Works with the OpenAI and Anthropic SDKs, LangChain, LiteLLM, Vercel AI SDK, Cursor, Claude Code and anything else that lets you set a base URL.

from openai import OpenAI

client = OpenAI(base_url="https://ai.dagentx.com/v1", api_key="crk_…")
r = client.chat.completions.create(
    model="venice/zai-org-glm-5",
    messages=[{"role": "user", "content": "Hello from Monad"}],
)
print(r.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://ai.dagentx.com/v1", apiKey: process.env.WRAITH_NEXUS_KEY });
const r = await client.chat.completions.create({
  model: "near/anthropic/claude-sonnet-5",
  messages: [{ role: "user", content: "Hello from Monad" }],
});
console.log(r.choices[0].message.content);

curl https://ai.dagentx.com/v1/chat/completions \
  -H "Authorization: Bearer $WRAITH_NEXUS_KEY" \
  -H "content-type: application/json" \
  -d '{"model":"venice/zai-org-glm-5","messages":[{"role":"user","content":"Hello from Monad"}]}'

How it works

Your wallet is your account.

Deposit USDC

Into CandorCredits on Monad, from any wallet. Your balance lives on chain, not in our database.

Register a key

Generated in your browser. Only its SHA-256 hash goes on chain. Revoke it from your wallet any time.

Call any model

Each request reserves its worst-case cost and bills what it actually used: provider price plus a 5% router fee.

Settle with proof

Usage settles on chain against a Merkle root of your signed call records. Check every line yourself.

Every modality

Chat and reasoning, vision, embeddings, image generation, text to speech and speech to text, all on the same key and balance.

Browse models and prices →

Private by default

Venice keeps no prompts or outputs; NEAR AI runs models in confidential compute. We store call records encrypted with post-quantum keys, sealed to your own key if you register one.

Built for agents

No sign-up for machines: answer a 402 with a USDC payment on Monad (x402) and get the response. Or prepay and sign in with the agent's wallet.

Agent guide →

Spending controls on chain

Set a daily cap or freeze your account from your wallet. The relay cannot settle more than you allow.

Verifiable

Records are signed with ML-DSA and anchored on Monad. Fetch a Merkle proof for any call and verify it in your browser.

Skills and SDKs

A ready-made skill for Claude Code and other agents, an llms.txt, and copy-paste setup for every major tool.

Downloads →

Need policies, SSO and your own audit stream?

Candor AI for enterprises runs the same gateway inside your cloud with Entra ID, DLP and compliance packs.

Candor AI Enterprise →