Chat Completions

Endpoint

http
POST https://api.aicodedog.com/v1/chat/completions

Supported models are listed at /models. Common picks: gpt-5.4, gpt-5.4-mini, claude-opus-4-1, claude-sonnet-4-6, DeepSeek-V4-Pro, Kimi-K2.6.

Minimal request

curl
curl https://api.aicodedog.com/v1/chat/completions \
  -H "Authorization: Bearer $AICODEDOG_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [
      {"role": "system", "content": "You are a concise assistant."},
      {"role": "user",   "content": "Explain RAG in one sentence."}
    ],
    "temperature": 0.7,
    "max_tokens": 500
  }'

Streaming (SSE)

Add "stream": true and the response becomes Server-Sent Events — every chunk is a small incremental JSON delta:

python
from openai import OpenAI
client = OpenAI(api_key="<KEY>", base_url="https://api.aicodedog.com/v1")

stream = client.chat.completions.create(
    model="gpt-5.4",
    messages=[{"role": "user", "content": "Tell me a 100-word joke"}],
    stream=True,
)
for chunk in stream:
    delta = chunk.choices[0].delta.content or ""
    print(delta, end="", flush=True)

node
const stream = await client.chat.completions.create({
  model: 'gpt-5.4',
  messages: [{ role: 'user', content: 'Tell me a 100-word joke' }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '');
}

Function calling / tools

Describe the tool with a JSON Schema. The model will return tool_calls when appropriate; execute the call yourself and feed the result back as the next-turn tool message:

python
tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Look up the current weather in a city",
        "parameters": {
            "type": "object",
            "properties": {"city": {"type": "string"}},
            "required": ["city"],
        },
    },
}]

resp = client.chat.completions.create(
    model="claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Is it hot in Beijing today?"}],
    tools=tools,
    tool_choice="auto",
)

tool_call = resp.choices[0].message.tool_calls[0]
# tool_call.function.name == "get_weather"
# tool_call.function.arguments == '{"city": "Beijing"}'

JSON mode

Forces the model to emit valid JSON. Use it for structured extraction or API integration:

json
{
  "model": "gpt-5.4",
  "messages": [
    {"role": "system", "content": "Convert user input into a {city, day} object."},
    {"role": "user",   "content": "What's the weather in Beijing on Tuesday?"}
  ],
  "response_format": { "type": "json_object" }
}

Multimodal (image input)

Vision-capable models (e.g. gpt-5.4, claude-opus-4-1, gemini-3-1-pro) accept mixed text + image input:

json
{
  "model": "gpt-5.4",
  "messages": [{
    "role": "user",
    "content": [
      {"type": "text", "text": "What does the sign say?"},
      {
        "type": "image_url",
        "image_url": { "url": "https://example.com/sign.jpg" }
      }
    ]
  }]
}

Base64 dataURLs are also supported (data:image/png;base64,iVBOR...) — handy for local files.

Full parameter table

• model (required) — model slug.
• messages (required) — conversation history, organised by system / user / assistant / tool roles.
• temperature — 0–2, default 1. 0 is near-deterministic, 2 is highly random.
• top_p — nucleus-sampling threshold, default 1. Pick this OR temperature, not both.
• max_tokens — output cap. Strongly recommended — set it to avoid runaway costs.
• stream — streaming output, true / false.
• stop — string or array — generation halts on encountering any of them.
• presence_penalty / frequency_penalty — -2.0–2.0, controls topic novelty / repetition.
• tools / tool_choice — function-calling configuration (see above).
• response_format — JSON mode (see above).
• seed — same seed + same input attempts to yield reproducible output (best-effort).
• user — your end-user identifier for abuse detection.

Response format

json
{
  "id": "chatcmpl-AbCdEfGhIjK",
  "object": "chat.completion",
  "created": 1747291200,
  "model": "gpt-5.4",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "RAG = Retrieval Augmented Generation: look up first, answer second."
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 24,
    "completion_tokens": 12,
    "total_tokens": 36
  }
}

finish_reason values: stop (natural end) / length (hit max_tokens) / tool_calls (tool triggered) / content_filter (blocked by safety policy).