Перейти к основному содержимому
OpenAI

GPT 6 Luna API

OpenAI

GPT 6 Luna входит в семейство моделей GPTСмотреть все модели GPT

OpenAI fastest and most cost-efficient GPT-6 model for high-volume and latency-sensitive work, with text and image input.

openai/gpt-6-luna

Документация

Диалог

Начните диалог

OpenAI fastest and most cost-efficient GPT-6 model for high-volume and latency-sensitive work, with text and image input.

Enter — отправить · Shift+Enter — новая строка

UnificAlly не хранит ваши промпты и ответы LLM. Мы сохраняем только данные об использовании токенов для биллинга.

Использует POST /v1/chat/completions с вашим API-ключом UnificAlly. Поддерживает системный и пользовательский промпты, инструменты, стриминг и рассуждение, где оно доступно.

Обновлено

What is the GPT 6 Luna API?

The GPT 6 Luna API is live on UnificAlly as openai/gpt-6-luna, at $0.015/1M tokens for input and $0.075/1M tokens for output. GPT 6 Luna is the small, fast model in OpenAI's GPT-6 family, released on September 22, 2026 alongside GPT 6 Sol. OpenAI calls it its fastest and most cost-efficient model yet. It takes text and image input and returns text, with a 1,050,000-token context window, a 128,000-token max output, and knowledge up to May 18, 2026. OpenAI's API price is half of GPT 5.6 Luna's, and it keeps the same reasoning effort range, from none up to max. It is built for the calls you make thousands of times a day, and at max effort it still gets close to Claude Opus 5 on agentic coding.

GPT 6 Luna API specs

Model IDopenai/gpt-6-luna
EndpointPOST https://direct-api.unifically.com/v1/chat/completions
Other formatsPOST /v1/responses, POST /v1/messages
Request typesChat completion, streaming, tool calls
InputText and images
OutputText, as synchronous JSON or server-sent events with "stream": true
Input rate$0.015/1M tokens
Output rate$0.075/1M tokens
Cache read rate$0.0015/1M tokens
Cache creation rate$0.01875/1M tokens

Billing is per token, with no subscription. These are the standard rates. A prompt over 272,000 tokens bills the whole request at the long-context rate. The OpenAI API pricing guide covers the rates across the GPT line.

How to use the GPT 6 Luna API

The GPT 6 Luna API uses /v1/chat/completions, the OpenAI-compatible endpoint shared by every language model on UnificAlly.

  1. Create an API key on the API keys page.
  2. Send a POST request to https://direct-api.unifically.com/v1/chat/completions with the model ID and a messages array.
  3. Read the reply from choices[0].message.content in the same response.
curl -X POST https://direct-api.unifically.com/v1/chat/completions   -H "Content-Type: application/json"   -H "Authorization: Bearer YOUR_API_KEY"   -d '{
    "model": "openai/gpt-6-luna",
    "messages": [
      {"role": "user", "content": "Classify this support ticket as billing, bug, or feature request: My card was charged twice."}
    ]
  }'

The response comes back synchronously with the completion:

{
  "id": "3f8e2a1c-7b4d-4e9a-9c2f-5d1e8b7a6c40",
  "object": "chat.completion",
  "created": 1710000000,
  "model": "openai/gpt-6-luna",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "billing"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 28,
    "completion_tokens": 64,
    "total_tokens": 92,
    "prompt_tokens_details": {
      "cached_tokens": 0
    },
    "completion_tokens_details": {
      "reasoning_tokens": 0
    }
  }
}

The usage object holds the token counts the request is billed on. Set "stream": true to receive tokens as they generate, and "reasoning_effort": "low" for the fastest replies. Send requests to direct-api.unifically.com so long replies finish in full.

GPT 6 Luna API parameters

ParameterTypeDefaultWhat it does
modelstringrequiredModel identifier. Use openai/gpt-6-luna
messagesarrayrequiredNon-empty array of message objects with role and content. Roles: system, developer, user, assistant, tool, function
streambooleanfalseStreams the response as server-sent events
stream_optionsobjectnoneSet { "include_usage": true } to get a final chunk with token usage. Only used when stream is true
temperaturenumbernoneSampling temperature between 0 and 2. Higher values produce more random output
max_tokensintegernoneMaximum number of tokens to generate
max_completion_tokensintegernoneAccepted as an alias for max_tokens
reasoning_effortstringnoneHow much the model reasons before answering
top_pnumbernoneNucleus sampling. An alternative to temperature
stopstring or arraynoneUp to four sequences where the model stops generating
toolsarraynoneTools the model may call. Each needs a type and a function definition
tool_choicestring or objectnonenone, auto, required, or a specific tool
response_formatobjectnoneSet { "type": "json_object" } for JSON mode

The GPT 6 Luna API documentation lists every supported endpoint and request format for this model.

Key features of GPT 6 Luna

Near-Opus coding at a fraction of the cost

On DeepSWE v1.1, Luna scores 66.6% at max effort for $0.22 a task. Claude Opus 5 at medium effort scores 68.9% for $3.29 and Claude Fable 5 65.4% for $6.09, so Luna costs 93% less per task than Opus 5 and 96% less than Fable 5. GPT 5.6 Luna scores 62.2%.

GPT 5.6 Luna's agent score for 94% less

On Agents' Last Exam, which runs long professional workflows across 55 fields, Luna scores 50.9% at max effort for $0.15 a task. GPT 5.6 Luna scores 50.4% on the same test and spends $2.57 a task to get there.

Fewer factual errors than GPT 5.6 Sol

On OpenAI's set of difficult factual prompts, Luna at max effort answers with a factual error 7.6% of the time, below GPT 5.6 Sol at 8.5% and GPT 5.6 Luna at 12.0%, for about $0.012 a prompt against $0.87 for GPT 5.6 Sol.

Built for throughput

Artificial Analysis measures about 132 output tokens per second at max effort and 143 at medium, and ranks Luna 6th of 174 models in its class on the Intelligence Index with a score of 37. A full run of that index costs about $0.07 per task, against $0.18 for GPT 5.6 Luna.

What GPT 6 Luna is best for

  • High-volume pipelines

    Classification, extraction, tagging, and routing at the lowest price in the GPT-6 family.

  • Summaries and short replies

    Chat frontends, support drafts, and digests where speed and cost decide the design.

  • Agent worker steps

    Bounded tool calls and subtasks while GPT 6 Sol or Astra does the planning.

  • Draft code and fixes

    66.6% on DeepSWE v1.1 at max effort, for $0.22 a task.

  • Vision at scale

    Screenshots and scanned pages read inline for high-volume document jobs.

  • Cached batch jobs

    A stable prompt template reused at the cache-read rate across the whole batch.

GPT 6 Luna use cases

Route the bulk of your traffic through Luna: ticket tagging, entity extraction, document summaries, and short assistant replies. Use it as the worker model inside an agent system, handling the many small tool calls while GPT 6 Sol plans and checks the result. Read screenshots, receipts, and scanned forms in bulk. For coding, use it for draft patches, quick explanations, and test scaffolding, where a near-Opus score at a few cents a task changes what is affordable to automate. Batch jobs that share one long prompt template get the most from caching, since cached reads bill at a tenth of the input rate.

GPT 6 Luna limitations

Luna is the small model and it shows on harder work. Artificial Analysis measures its Coding Agent Index at 41, two points below GPT 5.6 Luna, with lower SWE-Atlas scores, and its knowledge-work evals dropped about 75 Elo on GDPval and 45 on AA-Briefcase. It also spends more tokens per task than GPT 5.6 Luna, about 51,000 against 41,000 on that index. On the Intelligence Index it sits below Gemini 3.8 Flash and DeepSeek V4.1 Flash.

The strong benchmark numbers come from max effort, which means long thinking before the first answer token: Artificial Analysis measures about 109 seconds on its hard prompts. For latency-sensitive calls use low or medium effort and expect lower scores.

OpenAI rates Luna High for cybersecurity and biology risk, so it runs behind the same safeguards as GPT 5.6. OpenAI also says some of its jailbreak robustness reflects a broader tendency to refuse, including some legitimate requests.

Long prompts cost more. Once a request's prompt passes 272,000 tokens, the whole request bills at the long-context rate: double on input, cache read, and cache write, and 1.5x on output.

When to use GPT 6 Luna

Use Luna when volume and cost are the constraint: the calls you make thousands of times a day, the small agent steps, and the batch jobs where price per request decides whether the job is worth running. When quality on hard reasoning or coding starts to bind, step up to GPT 6 Sol, and keep GPT 6 Astra for the hardest tasks.

Вопросы и ответы

Частые вопросы о GPT 6 Luna

openai/gpt-6-luna, called through the OpenAI-compatible POST /v1/chat/completions endpoint with your UnificAlly API key.

Keys are made on the API keys page once you have an account, and each request sends one as Authorization: Bearer <key>. Point it at https://direct-api.unifically.com/v1/chat/completions with the model ID openai/gpt-6-luna; because the endpoint is OpenAI-compatible, an OpenAI SDK works once its base_url points at UnificAlly.

Post a messages array with "model": "openai/gpt-6-luna" to /v1/chat/completions and the answer arrives synchronously in choices[0].message.content, with token counts in usage. Add "stream": true for server-sent events, or "reasoning_effort": "low" for the fastest GPT 6 Luna API replies.

High-volume, cost-sensitive work: classification, extraction, summaries, routing, short replies, and the worker steps inside larger agent systems. OpenAI calls it its fastest and most cost-efficient model yet.

1,050,000 tokens, with a max output of 128,000 tokens and a knowledge cutoff of May 18, 2026. It takes text and image input and returns text.

For small tasks and draft patches, yes. It scores 66.6% on DeepSWE v1.1 at max effort for $0.22 a task, close to Claude Opus 5 at medium effort, 68.9% for $3.29. Artificial Analysis measures its Coding Agent Index at 41, two points below GPT 5.6 Luna, so step up to GPT 6 Sol for serious coding work. # price-ok: benchmark cost per task, not our rate

OpenAI's API price is half of GPT 5.6 Luna's. Scores are similar or better on OpenAI's evals, 50.9% against 50.4% on Agents' Last Exam and 7.6% against 12.0% factual error rate, at a much lower cost per task. Independent coding scores are slightly lower.

Yes: none, low, medium (the default), high, xhigh, and max. Keep it low for instant replies. The benchmark scores on this page come from max effort, which uses far more tokens.

Billing switches to the long-context rate for the whole request once the prompt passes 272,000 tokens. Input, cache read, and cache write bill at double the standard rate and output at 1.5x. Both rates are listed on the pricing page.

OpenAI rates Luna High for cybersecurity and biology risk, so it ships with the same safeguards as GPT 5.6. OpenAI also notes that some of its jailbreak robustness comes from a broader tendency to refuse, including some legitimate requests.

GPT 6 Luna: руководства и сравнения

Перед запуском модели ознакомьтесь с ценами, ограничениями и проверенными результатами.