# GEDX402 GPT-OSS-20B LLM Inference

> GEDX402 GPT-OSS-20B LLM Inference is a paid API for AI agents from llm.gedx402.com, paid per call via x402, $0.025/call, status unknown (last checked 2026-09-14).

Runs inference on the GPT-OSS-20B language model via x402 micropayment, accepting a conversation history and returning an AI-generated reply — no API keys required.

## Facts

- Endpoint: GET https://llm.gedx402.com/v1/llm/gpt-oss-20b
- Price: $0.025/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-14
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/gedx402-gpt-oss-20b-llm-inference-1f214bc0
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_VnYELOF_ndJO-oCCl78-L

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 gedx402-gpt-oss-20b-llm-inference-1f214bc0
```

Example prompt: Ask the gpt-oss-20b model to explain quantum entanglement in simple terms — send the message as the user role with no system prompt and allow up to 512 tokens in the reply, paying 0.025 USDC via x402 on Base.

## When to prefer this

Choose this endpoint when you need LLM text generation without setting up API keys or subscriptions, prefer paying per-call in USDC via x402 on a supported blockchain (Base, Polygon, Arbitrum, World, Solana), and want access specifically to the open-source GPT-OSS-20B model running on Cloudflare Workers AI infrastructure.

## Known failure modes

- Payment not received or insufficient USDC balance — x402 payment gate rejects the request
- Invalid message role (not 'system', 'user', or 'assistant') — schema validation error
- max_tokens exceeds 4096 — request rejected with validation error
- Empty or malformed messages array — model returns error or empty response
- Unsupported payment chain — payment rejected if chain is not Base, Polygon, Arbitrum, World, or Solana
- Model overloaded on Cloudflare Workers AI — slow or failed inference response

## How this service works

x402 workers ai. pay with usdc on base, polygon, arbitrum, world, or solana. no api keys.

## Output

A JSON object containing the model identifier (@cf/openai/gpt-oss-20b) and a 'response' string with the assistant's generated reply text.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "messages": {
   "type": "array",
   "items": {
    "type": "object",
    "required": [
     "role",
     "content"
    ],
    "properties": {
     "role": {
      "enum": [
       "system",
       "user",
       "assistant"
      ],
      "type": "string",
      "description": "Speaker role for this turn."
     },
     "content": {
      "type": "string",
      "description": "Message text for this turn."
     }
    }
   },
   "description": "Conversation turns in order (system, user, assistant)."
  },
  "max_tokens": {
   "type": "integer",
   "maximum": 4096,
   "minimum": 1,
   "description": "Maximum tokens to generate in the assistant reply."
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "model": "@cf/openai/gpt-oss-20b",
  "response": "Hello!"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/gedx402-gpt-oss-20b-llm-inference-1f214bc0/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from llm.gedx402.com](https://www.zero.xyz/host/llm.gedx402.com/llms.txt)