# Argo Hub — Deep LLM Inference (x402-paid)

> Argo Hub — Deep LLM Inference (x402-paid) is a paid API for AI agents from diomedes.tail1c6d58.ts.net, paid per call via x402, $0.05/call, status unknown (last checked 2026-09-18).

Runs a deep LLM inference call on a private GPU fleet (Qwen3-27B-class) with verifier-gated correctness guarantees, paid per-call in USDC via x402.

## Facts

- Endpoint: POST https://diomedes.tail1c6d58.ts.net/premium/deep
- Price: $0.05/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-18
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/argo-hub-deep-llm-inference-x402-paid-09707e1d
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_ULMjrvg5-FSSy9cFdPf0-

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 argo-hub-deep-llm-inference-x402-paid-09707e1d -d '<json body>'
```

Example prompt: Use the Argo Hub deep inference tier to answer this complex reasoning question: 'What are the key trade-offs between transformer and state-space models for long-context tasks?' — the endpoint should regenerate for free if the answer is wrong.

## When to prefer this

Choose this endpoint when you need high-quality, deep reasoning LLM output with a correctness guarantee, are comfortable with per-call USDC micropayments via x402 on Base, and want to avoid subscription-based APIs. Particularly valuable for agent pipelines that require reliable, verifiable answers and want automatic free regeneration on wrong outputs. Prefer over standard hosted APIs when cost-per-call economics matter and you want a self-hosted, private GPU backend rather than shared cloud inference.

## Known failure modes

- Payment not settled — x402 USDC transaction fails or wallet has insufficient funds, resulting in a 402 response and no inference
- Invalid input schema — missing required fields ('description', 'mimeType', 'input.method') returns a 400-class error
- Model timeout — long-running deep inference exceeds GPU time limits and returns a 504 or equivalent error
- Verifier loop — rare edge case where regenerated answers repeatedly fail correctness checks, potentially stalling the response
- Network unreachable — private Tailscale domain (ts.net) may be inaccessible outside the operator's network

## How this service works

x402-paid LLM APIs on a private GPU fleet (Qwen3.8-27B, VL 30B). USDC on Base, settle in seconds via a self-hosted facilitator. Unique: verifier-gated correctness on /premium/deep — wrong answers regenerated free. First call free, no wallet: POST /premium/first.

## Output

Returns a JSON object containing a 'text' field with the model's generated response and a 'model_tier' field set to 'deep', indicating which inference tier handled the request. If the verifier deems the answer incorrect, the response is regenerated at no additional cost.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "description",
  "mimeType",
  "input"
 ],
 "properties": {
  "input": {
   "type": "object",
   "required": [
    "method"
   ],
   "properties": {
    "body": {
     "type": "object"
    },
    "type": {
     "type": "string"
    },
    "method": {
     "type": "string"
    },
    "bodyType": {
     "type": "string"
    },
    "input_schema": {
     "type": "object"
    }
   }
  },
  "output": {
   "type": "object"
  },
  "mimeType": {
   "type": "string"
  },
  "description": {
   "type": "string"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "application/json",
 "example": {
  "text": "...",
  "model_tier": "deep"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/argo-hub-deep-llm-inference-x402-paid-09707e1d/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from diomedes.tail1c6d58.ts.net](https://www.zero.xyz/host/diomedes.tail1c6d58.ts.net/llms.txt)