LLM Developer Guide — API Docs & Connection Examples

API Connection Example

curl https://api.openai.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{
    "model": "gpt-5.5",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Explain quantum computing in one paragraph."}
    ],
    "temperature": 0.7,
    "max_tokens": 500,
    "stream": false
  }'

Set stream: true for real-time token streaming. Use temperature: 0 for deterministic responses. Requires an OpenAI API key from platform.openai.com/api-keys.

API Connection Example

curl https://api.anthropic.com/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2026-06-01" \
  -d '{
    "model": "claude-opus-4-8-20260501",
    "max_tokens": 1024,
    "system": "You are a knowledgeable assistant.",
    "messages": [
      {"role": "user", "content": "Explain reinforcement learning from human feedback."}
    ]
  }'

Anthropic uses a system parameter (not a message role). Enable "thinking": {"type": "enabled", "budget_tokens": 2000} for extended reasoning. Get your key at console.anthropic.com.

API Connection Example

curl https://generativelanguage.googleapis.com/v1beta/models/gemini-3.1-pro:generateContent?key=$GEMINI_API_KEY \
  -H "Content-Type: application/json" \
  -d '{
    "system_instruction": {
      "parts": [{"text": "You are a helpful data scientist."}]
    },
    "contents": [
      {"role": "user", "parts": [{"text": "Explain the transformer architecture."}]}
    ],
    "generationConfig": {
      "temperature": 0.7,
      "maxOutputTokens": 800
    }
  }'

Gemini uses a RESTful generateContent endpoint. Pass the API key as a query parameter. Multimodal requests add inlineData parts for images. Get a free key at aistudio.google.com.

API Connection Example

curl https://api.anthropic.com/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2026-06-01" \
  -d '{
    "model": "claude-sonnet-4-6-20260501",
    "max_tokens": 1024,
    "system": "You are a senior software engineer.",
    "messages": [
      {"role": "user", "content": "Review this Python code for security issues."}
    ]
  }'

Same API structure as Opus — swap model to claude-sonnet-4-6-20260501. For production, enable prompt caching on system messages to cut costs.

API Connection Example

curl https://api.openai.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{
    "model": "gpt-5.4",
    "messages": [
      {"role": "system", "content": "You are a content editor."},
      {"role": "user", "content": "Summarize this article in 3 bullet points."}
    ],
    "temperature": 0.5,
    "max_tokens": 400
  }'

Identical API surface to GPT-5.5 — just change the model name. Production tip: use response_format: {"type": "json_object"} for structured extraction.

API Connection Example

curl https://api.deepseek.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $DEEPSEEK_API_KEY" \
  -d '{
    "model": "deepseek-chat",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Write a Python function to merge two sorted lists."}
    ],
    "temperature": 0.0,
    "max_tokens": 800,
    "stream": false
  }'

Drop-in OpenAI replacement — swap the base URL and model name. DeepSeek's deep reasoning mode is available with "model": "deepseek-reasoner".

API Connection Example

curl https://generativelanguage.googleapis.com/v1beta/models/gemini-3.5-flash:generateContent?key=$GEMINI_API_KEY \
  -H "Content-Type: application/json" \
  -d '{
    "system_instruction": {
      "parts": [{"text": "You classify support tickets by urgency."}]
    },
    "contents": [
      {"role": "user", "parts": [{"text": "My website is completely down — customers can\\'t check out!"}]}
    ],
    "generationConfig": {
      "temperature": 0.0,
      "maxOutputTokens": 100,
      "responseMimeType": "application/json",
      "responseSchema": {
        "type": "object",
        "properties": {
          "category": {"type": "string", "enum": ["urgent","high","medium","low"]},
          "summary": {"type": "string"}
        }
      }
    }
  }'

With sub-200ms median latency, Flash is ideal for real-time classification. Use responseSchema for structured JSON output — no parsing hacks needed.

API Connection Example

curl https://api.moonshot.cn/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $MOONSHOT_API_KEY" \
  -d '{
    "model": "moonshot-v1-128k",
    "messages": [
      {"role": "system", "content": "You are a legal document analyst."},
      {"role": "user", "content": "Review this 200-page contract and identify all indemnification clauses."}
    ],
    "temperature": 0.3,
    "max_tokens": 2000
  }'

Drop-in OpenAI replacement. For files >1MB, use the file upload endpoint first, then pass file_ids in the request. Get a key at platform.moonshot.cn.

API Connection Example

curl https://open.bigmodel.cn/api/paas/v4/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $ZHIPU_API_KEY" \
  -d '{
    "model": "glm-5.1",
    "messages": [
      {"role": "system", "content": "You are a bilingual translator and cultural consultant."},
      {"role": "user", "content": "Translate this marketing copy into Chinese, preserving the brand tone."}
    ],
    "temperature": 0.6,
    "max_tokens": 1000
  }'

OpenAI-compatible endpoint. Add "tools" for web search and "tool_choice" to force function calls. API key from open.bigmodel.cn.

API Connection Example (Ollama — Self-Hosted)

# 1. Download and run Gemma 4 locally (no API key needed)
ollama pull gemma4:27b

# 2. Chat Completions endpoint (OpenAI-compatible, localhost)
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemma4:27b",
    "messages": [
      {"role": "system", "content": "You are an offline privacy-first assistant."},
      {"role": "user", "content": "Summarize this document locally — no data leaves this machine."}
    ],
    "temperature": 0.3,
    "max_tokens": 600,
    "stream": false
  }'

No API key. No internet required. Runs entirely on your hardware. For production, wrap Ollama in a Docker container on Google Cloud Run or any VPS with a GPU.