Chat completions

POST /v1/chat/completions — the OpenAI chat schema across the AIx chat catalog, with health-aware route selection and automatic failover.

Request

curl https://api.aix.theaimart.co/v1/chat/completions \
  -H "Authorization: Bearer $AIX_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-sonnet-4-6",
    "messages": [
      { "role": "system", "content": "You are a senior Rust engineer." },
      { "role": "user", "content": "Make this function async." }
    ],
    "temperature": 0.4,
    "max_tokens": 1024
  }'

r = client.chat.completions.create(
    model="anthropic/claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Make this function async."}],
    temperature=0.4,
)
print(r.choices[0].message.content)

const r = await client.chat.completions.create({
  model: "anthropic/claude-sonnet-4-6",
  messages: [{ role: "user", content: "Make this function async." }],
  temperature: 0.4,
});
console.log(r.choices[0].message.content);

Try it

Try it Runs a live request with your key — stored only in this browser.

API key Model

Prompt

Parameters

Field	Type	Notes
`model`	string	Required. Canonical model id.
`messages`	array	Required. `{role, content}`; roles: `system` / `user` / `assistant`.
`stream`	bool	SSE token streaming — see Streaming.
`temperature`	number	0–2. Default `0.7`.
`max_tokens`	int	Max output tokens. Default `256`.
`top_p`	number	Nucleus sampling. Default `1.0`.
`stop`	string \| array	Stop sequence(s).
`presence_penalty`	number	−2 to 2.
`frequency_penalty`	number	−2 to 2.

Response

{
  "id": "chatcmpl-…",
  "object": "chat.completion",
  "model": "anthropic/claude-sonnet-4-6",
  "choices": [{
    "index": 0,
    "message": { "role": "assistant", "content": "…" },
    "finish_reason": "stop"
  }],
  "usage": { "prompt_tokens": 41, "completion_tokens": 128, "total_tokens": 169 }
}

The response matches the OpenAI chat.completion shape. Tool calls and structured-output fields are preserved when supported by the chosen model.

Billing

We deduct an estimate up front, then reconcile down to the reported token usage. See Billing.

Prompts are data, not code to sanitize. Backticks, $, angle brackets and full code blocks pass through untouched — built for coding workloads.

← Models

Streaming →