# Xona Prova Verified Inference (TEE-Attested Chat Completions)

> Xona Prova Verified Inference (TEE-Attested Chat Completions) is a paid API for AI agents from api.xona-agent.com, paid per call via x402, $0.0014/call, status unknown (last checked 2026-10-01).

Runs LLM chat completions inside a Trusted Execution Environment (TEE) with hardware attestation and anchors each response as an on-chain proof on Solana, returning a Solscan link.

## Facts

- Endpoint: POST https://api.xona-agent.com/llm/prova?utm_source=zero.xyz
- Price: $0.0014/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-01
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/xona-prova-verified-inference-tee-attested-chat-completions-614fdc26
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_k-7TWSOEOCGil3dwSOA7r

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 xona-prova-verified-inference-tee-attested-chat-completions-614fdc26 -d '<json body>'
```

Example prompt: Using the Prova verified inference endpoint, run deepseek-v4-pro with a temperature of 0.7 and max 512 tokens to answer: 'Summarize the key risks of smart contract auditing' — I need the Solana on-chain proof link alongside the response.

## When to prefer this

Choose this endpoint when verifiability, auditability, or tamper-proof proof of AI inference matters — e.g. compliance workflows, AI-generated legal or financial content, or any use case where you need to prove the exact model and output via a Solana on-chain record. Not needed for standard LLM calls where attestation is irrelevant.

## Known failure modes

- Invalid or unsupported model ID returns 400 with list of valid models
- Payment failure or insufficient USDC returns 402 Payment Required
- TEE attestation failure returns 500 with attestation error detail
- Prompt or messages missing returns 422 validation error
- Token limit exceeded returns 400 with max_tokens constraint
- Model tier unavailable returns 503 with retry guidance

## How this service works

Verified inference (Prova). OpenAI-compatible chat completions run inside a TEE (NVIDIA + Intel), hardware-attested, and anchored on Solana — every response returns an on-chain proof + Solscan link. Select a model via the body `model` field; price varies by tier (standard/premium/flagship).

## Output

An OpenAI-compatible chat completion response plus a hardware attestation proof and a Solscan link confirming the response was anchored on Solana, providing cryptographic verifiability of the inference.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "model": {
   "type": "string",
   "default": "deepseek-v4-flash",
   "description": "Model id (price varies by tier). One of: deepseek-v4-flash, gpt-oss-20b, qwen-2.5-7b, qwen3.5-27b, qwen3.6-27b, qwen3-vl-30b, gpt-oss-120b, gemma-4-31b, qwen3.5-122b, kimi-k2.6, glm-5.2, deepseek-v4-pro"
  },
  "prompt": {
   "type": "string",
   "description": "The user prompt (or use messages)"
  },
  "messages": {
   "type": "array",
   "items": {
    "type": "string"
   },
   "description": "Optional OpenAI-style messages array (overrides prompt/system_prompt)"
  },
  "max_tokens": {
   "type": "number",
   "description": "Optional cap on output tokens"
  },
  "temperature": {
   "type": "number",
   "description": "Optional sampling temperature"
  },
  "system_prompt": {
   "type": "string",
   "description": "Optional system prompt"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/xona-prova-verified-inference-tee-attested-chat-completions-614fdc26/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from api.xona-agent.com](https://www.zero.xyz/host/api.xona-agent.com/llms.txt)