# Combify Compute Chat

> Combify Compute Chat is a paid API for AI agents from combify-compute-production.up.railway.app, paid per call via x402, $0.01/call, status unknown (last checked 2026-10-29).

Runs a user prompt through Llama 3.1 8B Instruct and returns generated text for a fixed $0.01 USDC payment per call

## Facts

- Endpoint: POST https://combify-compute-production.up.railway.app/chat?utm_source=zero.xyz
- Price: $0.01/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-29
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/combify-compute-chat-fb2607db
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_lIYow8b5PMvuuP578kJh1

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 combify-compute-chat-fb2607db -d '<json body>'
```

Example prompt: Ask the Llama 3.1 8B model to explain the difference between supervised and unsupervised learning in plain English, keeping the answer under 200 tokens.

## When to prefer this

Choose this endpoint when you need low-cost, on-demand text generation from a capable small LLM (Llama 3.1 8B) and want predictable per-call pricing in USDC via the x402 payment protocol. It is well-suited for agents that need to run many short inference calls economically without a subscription, or for workflows that require pay-as-you-go LLM access settled on-chain. Prefer larger hosted LLMs (e.g. GPT-4, Claude) for complex reasoning tasks requiring larger context or superior accuracy.

## Known failure modes

- 402 Payment Required — request was made without a valid USDC payment header or insufficient funds
- prompt exceeds 2000 character limit — request rejected with validation error
- max_tokens out of range (less than 1 or greater than 512) — validation error
- model inference timeout or Railway infrastructure error — 5xx response
- network connectivity issues to railway.app deployment

## How this service works

Short prompt in, text out (Llama 3.1 8B instruct), fixed price $0.01 USDC. POST JSON body {prompt (required, max 2000 chars), max_tokens (optional 1-512, default 256)}. Unpaid calls get 402; paid calls settle only on success.

## Output

A plain text string containing the model's generated response, limited to the number of tokens specified (default 256, max 512). The response is only returned and payment settled if the call succeeds; unpaid requests receive a 402 error.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "prompt": {
   "type": "string",
   "description": "User prompt, max 2000 chars"
  },
  "max_tokens": {
   "type": "integer",
   "description": "Max output tokens, 1 to 512"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/combify-compute-chat-fb2607db/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from combify-compute-production.up.railway.app](https://www.zero.xyz/host/combify-compute-production.up.railway.app/llms.txt)