OpenAI: GPT-5.6 Luna

GPT-5.6 Luna is the lightweight model in OpenAI's GPT-5.6 series. It is designed for cost-efficient reasoning, coding, and agentic workflows where throughput and responsiveness matter more than using the highest-capability tier.

WaveSpeed AI exposes openai/gpt-5.6-luna through an OpenAI-compatible API, so it can be used with standard OpenAI SDKs and existing chat-completions-based application flows.

Why Use GPT-5.6 Luna

• Lightweight GPT-5.6 model for high-volume workloads
• Cost-efficient reasoning and coding for large-scale production use
• Well suited for agentic workflows, tool use, and structured outputs
• Supports long-context workloads such as document analysis and extended multi-turn sessions
• Works with both Chat Completions and Responses API workflows

Key Features

• Context Window: 1,050,000 tokens
• Max Output: 128,000 tokens
• Vision Input: Supported
• Function Calling: Supported
• Structured Outputs: Supported
• Best Fit: cost-sensitive reasoning, coding, agents, high-throughput tasks

Specifications

Pricing

Standard rates apply when the request contains no more than 272,000 input tokens.

Long-context pricing

Requests with 272,001 or more input tokens are billed at the following rates for the full request:

Cache writes are billed at 1.25× the applicable uncached input rate. Tool charges, including Web Search, are billed separately.

How to Use

Chat Completions

Use Chat Completions when you want a straightforward OpenAI-compatible integration path for standard conversational and coding workflows.

Python

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://llm.wavespeed.ai/v1"
)

response = client.chat.completions.create(
    model="openai/gpt-5.6-luna",
    messages=[
        {"role": "user", "content": "Summarize this issue in one paragraph."}
    ]
)

print(response.choices[0].message.content)

cURL

curl https://llm.wavespeed.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "openai/gpt-5.6-luna",
    "messages": [
      {"role": "user", "content": "Summarize this issue in one paragraph."}
    ]
  }'

Pro Mode

GPT-5.6 Luna supports a stronger reasoning path through Pro mode.

Pro mode is not a separate core model that you need to configure independently. Instead, use the same base model, openai/gpt-5.6-luna, and enable Pro mode in the Responses API with:

{
  "reasoning": {
    "mode": "pro"
  }
}

Use Pro mode when you want the model to spend more effort on difficult reasoning, planning, and tool-using tasks. It is a better fit for complex coding, high-stakes decision logic, and multi-step agent workflows where answer quality matters more than speed or token efficiency.

In practice, Pro mode usually means:

• higher reasoning depth
• better consistency on difficult tasks
• more token usage
• higher latency than standard requests

Responses API Example

Python

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://llm.wavespeed.ai/v1"
)

response = client.responses.create(
    model="openai/gpt-5.6-luna",
    input="Review this automation design and identify the main reliability risk.",
    reasoning={
        "mode": "pro",
        "effort": "medium"
    }
)

print(response.output_text)

cURL

curl https://llm.wavespeed.ai/v1/responses \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "openai/gpt-5.6-luna",
    "input": "Review this automation design and identify the main reliability risk.",
    "reasoning": {
      "mode": "pro",
      "effort": "medium"
    }
  }'

When to Use Pro Mode

Choose Pro mode for:

• multi-step coding and debugging
• agent workflows with tools or long chains of reasoning
• tasks that require deeper analysis instead of quick turnaround
• prompts where higher accuracy is worth additional cost and latency

Use standard mode when:

• latency matters more than depth
• the task is simple or repetitive
• you are optimizing for throughput or cost

Notes

• Designed as the most cost-efficient option within the GPT-5.6 family
• Best paired with the Responses API when reasoning, tool use, or multi-turn state matters