# AgentBit LLM Completion

> AgentBit LLM Completion is a paid API for AI agents from agentbit.app, paid per call via x402, $0.008/call, status unknown (last checked 2026-10-02).

Send a text prompt (with optional system role) and receive an LLM-generated answer for reasoning, drafting, classification, or rewriting tasks.

## Facts

- Endpoint: POST https://agentbit.app/v1/ai/ask?utm_source=zero.xyz
- Price: $0.008/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-02
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/agentbit-llm-completion-d0b4f5e8
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_qtgXmE6HDeTe_FrLBmCnt

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 agentbit-llm-completion-d0b4f5e8 -d '<json body>'
```

Example prompt: Ask the LLM: 'You are a formal business writing assistant' — then complete this prompt with max 800 tokens: 'Rewrite the following customer complaint email in a calm, professional tone: [email text here].'

## When to prefer this

Choose this endpoint when you need a flat-rate, pay-per-call LLM completion without managing API keys, billing accounts, or model selection. Ideal for agents that need occasional text generation, classification, or reasoning steps billed in USDC microtransactions. Best when the task fits within the 1024-token output cap and the user does not need streaming or fine-tuned model selection.

## Known failure modes

- Prompt exceeds ~16 KB input limit — request rejected
- max_tokens exceeds hard cap of 1024 — capped or rejected
- Temperature out of 0–2 range — validation error
- Payment not included or insufficient — 402 Payment Required
- LLM returns truncated output if max_tokens is too low for the task
- Ambiguous or malformed prompt may produce low-quality completion

## How this service works

General LLM completion: send a prompt (optional system role), get an answer. Reason, rewrite, classify, draft or answer. Flat price, input/output capped.

## Output

Returns a generated text string that is the LLM's completion of the submitted prompt. The response is shaped by the optional system instruction, capped at the specified max_tokens (default 512, hard limit 1024), and sampled at the given temperature. Useful for answers, rewrites, classifications, drafts, or reasoning outputs.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "prompt": {
   "type": "string",
   "description": "The user prompt (required, up to ~16 KB)."
  },
  "system": {
   "type": "string",
   "description": "Optional system instruction to steer style/role."
  },
  "max_tokens": {
   "type": "integer",
   "description": "Max output tokens (default 512, hard cap 1024)."
  },
  "temperature": {
   "type": "number",
   "description": "Sampling temperature 0-2 (default 0.3)."
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/agentbit-llm-completion-d0b4f5e8/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from agentbit.app](https://www.zero.xyz/host/agentbit.app/llms.txt)