Abubuwan da aka Tsara

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"]
        }
      }
    }
  }'

JSON shi ne rubutun amsar, a cikin 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}
  }
}

Yanayin response_format

response_format field ne na /v1/chat/completions. Ba tare da shi ba amsar rubutu ce ta yau da kullun.

Yanayin JSON object ba ya buƙatar schema. Ku ambaci keys a cikin 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"}
}

Tallafi ga kowane model

GET /v1/models yana bayar da rahoton iri ɗaya ga kowane id: capabilities.json_schema a kowane model, da capabilities.response_format a hosted open-weight models. Models & farashi

Keywords na schema da ake tallafawa

Abin da model ke karɓa daga schema ɗinku ya dogara da model da kuma endpoint:

Ƙaramin saitin keywords na models na Shannon a /v1/chat/completions:

A /v1/responses: text.format

/v1/responses yana rubuta request iri ɗaya a matsayin text.format. Schema yana kai tsaye a cikin format, kusa da name da strict, ba a ƙarƙashin key na json_schema ba. Don yanayin JSON object ku aika {"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"]
      }
    }
  }
}

JSON shi ne text na ɓangaren output_text na item ɗin message:

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

Tallafi ga kowane model da kurakuran iri ɗaya ne da na /v1/chat/completions. Responses API

A /v1/messages: tool a matsayin schema

Tsarin Messages ba shi da field na structured-output. Don JSON daga models iri ɗaya, ku kira /v1/chat/completions ko /v1/responses da maɓalli ɗaya.

Don ku tsaya a /v1/messages, ku bayyana object ɗin a matsayin tool: ku sa schema ɗinku a cikin input_schema kuma ku ambaci sunan tool a cikin tool_choice. input na block ɗin tool_use sai ya zama object ɗinku, an riga an yi parse. Hosted open-weight models suna aiwatar da tool_choice mai suna.

{
  "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"}
    }
  ]
}

Ku tabbatar stop_reason shi ne tool_use kafin ku karanta block ɗin. Kiran Aiki

Yadda JSON yake zuwa

• JSON string ne. Amsar ba ta da object da aka yi parse kusa da ita: ku gudanar da JSON parser ɗinku a kan message.content.
• Da stream: true JSON yana zuwa a guntayen delta.content. Ku haɗa su kuma ku yi parse bayan stream ya ƙare; guntu shi kaɗai ba JSON mai inganci ba ne.
• Hosted open-weight model yana amsa request na response_format ba tare da rubutun reasoning ba: reasoning_content shi ne null.

Ku duba abin da kuka karɓa

Schema yana jagorantar yadda model ke rubutu. API ba ya kwatanta amsar da ta kammala da schema ɗinku, don haka code ɗinku ne ke yin binciken ƙarshe:

• Ku yi parse a cikin error handler, kamar yadda samfurin da ke sama ya yi.
• Ku tabbatar da ingancin object da aka yi parse da schema iri ɗaya a cikin code ɗinku, misali da JSON Schema library, Pydantic ko Zod.
• Amsa mai tsawo na iya ƙarewa kafin JSON ya cika. Amsar da ba streaming ba tana bayar da finish_reason a matsayin stop ko tool_calls kawai, don haka kuskuren parse shi ne alamar. Ku sake nema, ko ku nemi ƙaramin object.

Kurakurai

Ana aika amsoshin 400 biyun kafin a cire komai daga balance ɗinku. Kuskuren Gudanarwa