Gestruktureerde uitvoer

POST https://api.shannon-ai.com/v1/chat/completions

import json
from openai import OpenAI

client = OpenAI(api_key="YOUR_API_KEY", base_url="https://api.shannon-ai.com/v1")

response = client.chat.completions.create(
    model="Kimi-K3-3BIT-REAP",
    messages=[
        {"role": "user", "content": "Extract the person: John Doe, 30 years old, engineer."}
    ],
    response_format={
        "type": "json_schema",
        "json_schema": {
            "name": "person",
            "strict": True,
            "schema": {
                "type": "object",
                "properties": {
                    "name": {"type": "string"},
                    "age": {"type": "integer"},
                    "occupation": {"type": "string"},
                },
                "required": ["name", "age", "occupation"],
            },
        },
    },
)

# The JSON arrives as text in message.content: parse it, then check it.
try:
    person = json.loads(response.choices[0].message.content)
except json.JSONDecodeError:
    person = None  # ask again, or report the failure

print(person)

import OpenAI from "openai";

const client = new OpenAI({ apiKey: "YOUR_API_KEY", baseURL: "https://api.shannon-ai.com/v1" });

const response = await client.chat.completions.create({
  model: "Kimi-K3-3BIT-REAP",
  messages: [
    { role: "user", content: "Extract the person: John Doe, 30 years old, engineer." },
  ],
  response_format: {
    type: "json_schema",
    json_schema: {
      name: "person",
      strict: true,
      schema: {
        type: "object",
        properties: {
          name: { type: "string" },
          age: { type: "integer" },
          occupation: { type: "string" },
        },
        required: ["name", "age", "occupation"],
      },
    },
  },
});

// The JSON arrives as text in message.content: parse it, then check it.
let person = null;
try {
  person = JSON.parse(response.choices[0].message.content);
} catch {
  // ask again, or report the failure
}

console.log(person);

curl https://api.shannon-ai.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Kimi-K3-3BIT-REAP",
    "messages": [
      {"role": "user", "content": "Extract the person: John Doe, 30 years old, engineer."}
    ],
    "response_format": {
      "type": "json_schema",
      "json_schema": {
        "name": "person",
        "strict": true,
        "schema": {
          "type": "object",
          "properties": {
            "name": {"type": "string"},
            "age": {"type": "integer"},
            "occupation": {"type": "string"}
          },
          "required": ["name", "age", "occupation"]
        }
      }
    }
  }'

Die JSON is die teks van die antwoord, in choices[0].message.content:

{
  "id": "chatcmpl-0c4e7a1b9d2f4e6a8b3c5d7e9f1a2b3c",
  "object": "chat.completion",
  "created": 1791590400,
  "model": "Kimi-K3-3BIT-REAP",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "{\"name\":\"John Doe\",\"age\":30,\"occupation\":\"engineer\"}",
        "reasoning_content": null
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 96,
    "completion_tokens": 19,
    "total_tokens": 115,
    "prompt_tokens_details": {"cached_tokens": 0},
    "completion_tokens_details": {"reasoning_tokens": 0}
  }
}

Die modusse van response_format

response_format is 'n veld van /v1/chat/completions. Daarsonder is die antwoord gewone teks.

JSON-objekmodus het geen skema nodig nie. Noem die sleutels in die prompt:

{
  "model": "shannon-3",
  "messages": [
    {
      "role": "user",
      "content": "Name the capital of France. Answer as JSON with the keys city and country."
    }
  ],
  "response_format": {"type": "json_object"}
}

Ondersteuning per model

GET /v1/models meld dieselfde per id: capabilities.json_schema op elke model, en capabilities.response_format op die gehuisveste oopgewig-modelle. Modelle en pryse

Ondersteunde skema-sleutelwoorde

Wat 'n model van jou skema ontvang, hang van die model en van die eindpunt af:

Die sleutelwoord-substel van die Shannon-modelle op /v1/chat/completions:

Op /v1/responses: text.format

/v1/responses skryf dieselfde versoek as text.format. Die skema sit direk in format, langs name en strict, nie onder 'n json_schema-sleutel nie. Vir JSON-objekmodus stuur {"format": {"type": "json_object"}}.

{
  "model": "Kimi-K3-3BIT-REAP",
  "input": "Extract the person: John Doe, 30 years old, engineer.",
  "text": {
    "format": {
      "type": "json_schema",
      "name": "person",
      "strict": true,
      "schema": {
        "type": "object",
        "properties": {
          "name": {"type": "string"},
          "age": {"type": "integer"},
          "occupation": {"type": "string"}
        },
        "required": ["name", "age", "occupation"]
      }
    }
  }
}

Die JSON is die text van die output_text-deel van die message-item:

{
  "output": [
    {
      "id": "msg_7e1d3c5b9a2f4d6e8c0b1a3f5e7d9c2b",
      "type": "message",
      "role": "assistant",
      "status": "completed",
      "content": [
        {
          "type": "output_text",
          "text": "{\"name\":\"John Doe\",\"age\":30,\"occupation\":\"engineer\"}",
          "annotations": []
        }
      ]
    }
  ]
}

Ondersteuning per model en die foute is dieselfde as op /v1/chat/completions. Responses API

Op /v1/messages: 'n gereedskap as die skema

Die Messages-formaat het geen veld vir gestruktureerde uitvoer nie. Vir JSON van dieselfde modelle, roep /v1/chat/completions of /v1/responses met dieselfde sleutel op.

Om op /v1/messages te bly, beskryf die objek as 'n gereedskap: sit jou skema in input_schema en noem die gereedskap in tool_choice. Die input van die tool_use-blok is dan jou objek, reeds ontleed. 'n Benoemde tool_choice word deur die gehuisveste oopgewig-modelle toegepas.

{
  "model": "Kimi-K3-3BIT-REAP",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": "Extract the person: John Doe, 30 years old, engineer."
    }
  ],
  "tools": [
    {
      "name": "record_person",
      "description": "Record the extracted person.",
      "input_schema": {
        "type": "object",
        "properties": {
          "name": {"type": "string"},
          "age": {"type": "integer"},
          "occupation": {"type": "string"}
        },
        "required": ["name", "age", "occupation"]
      }
    }
  ],
  "tool_choice": {"type": "tool", "name": "record_person"}
}

{
  "stop_reason": "tool_use",
  "content": [
    {
      "type": "tool_use",
      "id": "toolu_4b8d2f6a1c3e4a5b9d7f0e2c4a6b8d1f",
      "name": "record_person",
      "input": {"name": "John Doe", "age": 30, "occupation": "engineer"}
    }
  ]
}

Toets dat stop_reason tool_use is voordat jy die blok lees. Funksie-oproepe

Hoe die JSON aankom

• Die JSON is 'n string. Die antwoord het nie 'n ontlede objek daarnaas nie: laat jou JSON-ontleder oor message.content loop.
• Met stream: true kom die JSON in delta.content-stukke aan. Voeg hulle saam en ontleed een keer nadat die stroom geëindig het; 'n stuk op sy eie is nie geldige JSON nie.
• 'n Gehuisveste oopgewig-model beantwoord 'n response_format-versoek sonder redeneerteks: reasoning_content is null.

Toets wat jy ontvang

Die skema lei hoe die model skryf. Die API vergelyk nie die voltooide antwoord met jou skema nie, so jou kode doen die finale toets:

• Ontleed binne 'n foutafhandelaar, soos die voorbeeld heel bo doen.
• Valideer die ontlede objek met dieselfde skema in jou eie kode, byvoorbeeld met 'n JSON Schema-biblioteek, Pydantic of Zod.
• 'n Lang antwoord kan eindig voordat die JSON volledig is. 'n Antwoord wat nie gestroom is nie, meld finish_reason net as stop of tool_calls, so 'n ontlede-fout is die sein. Vra weer, of vra vir 'n kleiner objek.

Foute

Die twee 400-antwoorde word gestuur voordat enigiets van jou balans afgetrek word. Fouthantering