# NVIDIA Nemotron-3 Ultra 550B via x402 Micropayments

> NVIDIA Nemotron-3 Ultra 550B via x402 Micropayments is a paid API for AI agents from x402-nvidia.vercel.app, paid per call via x402, $0.129626/call, status unknown (last checked 2026-09-13).

Run inference on NVIDIA's Nemotron-3 Ultra 550B large language model via pay-per-call USDC micropayments using the x402 protocol

## Facts

- Endpoint: POST https://x402-nvidia.vercel.app/api/nemotron-3-ultra-550b
- Price: $0.129626/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-13
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/nvidia-nemotron-3-ultra-550b-via-x402-micropayments-c2a80acb
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_IffbiUAw7nstF_1mSz71N

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability nvidia-nemotron-3-ultra-550b-via-x402-micropayments-c2a80acb -d '<json body>'
```

Example prompt: Use the NVIDIA Nemotron-3 Ultra 550B model to answer this question with detailed reasoning: 'What are the key tradeoffs between transformer and state-space architectures for long-context language modeling?'

## When to prefer this

Choose this endpoint when you need a very large (550B parameter) NVIDIA language model for complex reasoning, nuanced generation, or high-quality completions, and want to pay per call in USDC via x402 without a subscription. Prefer over smaller models when task complexity demands frontier-scale reasoning capacity.

## Known failure modes

- Payment failure if x402 USDC micropayment is rejected or wallet has insufficient funds
- 402 Payment Required if x402 header is missing or malformed
- Model unavailable or timeout for very long prompts against the 550B model
- Malformed request body returns 400 if messages array is missing or improperly structured
- Rate limiting if too many concurrent requests are sent

## How this service works

Access 48+ NVIDIA NIM AI models via x402 micropayments. Chat completions, vision, safety, translation, and more.

## Output

Returns a chat completion object with an array of choices, each containing an assistant message with the model's generated response content. Structured as a standard OpenAI-compatible chat completion response.

## Response schema (JSON Schema)

```json
{
 "type": "object",
 "example": {
  "choices": [
   {
    "message": {
     "role": "assistant",
     "content": "Hello!"
    }
   }
  ]
 },
 "properties": {
  "choices": {
   "type": "array",
   "description": "Model responses"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/nvidia-nemotron-3-ultra-550b-via-x402-micropayments-c2a80acb/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from x402-nvidia.vercel.app](https://www.zero.xyz/host/x402-nvidia.vercel.app/llms.txt)