What is the GPT-5.6 Luna context window?

1,050,000 tokens, with a maximum output of 128,000 tokens — the same as every other model in the GPT-5.6 family. The cheaper tier does not come with a smaller window, which is the detail that makes it usable on long inputs.

Can I turn reasoning off on GPT-5.6 Luna?

Yes. Effort accepts none, which OpenAI describes for latency-critical tasks that do not benefit from reasoning or multi-chained tool calls — voice, fast retrieval and classification are the named cases. Note the family default is medium, so a request that omits the field is paying for reasoning you may not want on this tier.

What reasoning effort levels does GPT-5.6 Luna accept?

GPT-5.6 Luna accepts six: none, low, medium, high, xhigh and max. There is no minimal on this family. Every rung really is distinct, unlike some models that accept a long list and quietly collapse it, so a middle value buys a middle cost.

Does GPT-5.6 Luna read images?

Yes. Input is text and image, output is text. This generation accepts images at their original dimensions with original or automatic detail, so a screenshot does not need resizing before it goes in.

What is the GPT-5.6 Luna knowledge cutoff?

OpenAI lists February 16, 2026 for the whole GPT-5.6 family. For anything more recent, give GPT-5.6 Luna the material in the request or wire up a search tool.

Which endpoint and model string does GPT-5.6 Luna use on reAPI?

An OpenAI-compatible `/v1/chat/completions` endpoint on api.reapi.ai, with `gpt-5.6-luna` as the model string. OpenAI notes that its reasoning models work better on the newer Responses API while Chat Completions remains supported — the family's changelog lists both endpoints.

When should I use GPT-5.6 Terra instead?

When the task involves planning, judgement or a real tool loop. Terra is the balanced tier and sits at the family's default effort comfortably; GPT-5.6 Luna is where you go when volume and latency dominate and the work is closer to sorting than to deciding. The versus table on this page compares them row by row.

Is GPT-5.6 Luna free or paid?

Paid. Usage is billed pay-as-you-go in USD from your api.reapi.ai token balance, based on input, output, and any applicable cache-token rates. See the live per-1M-token rates in the pricing table on this page; failed requests are not charged.

Does GPT-5.6 Luna require a monthly subscription?

No. There is no monthly subscription. Add funds to your api.reapi.ai token balance and pay in USD only for the input, output, and any applicable cache tokens each request uses. The live per-1M-token rates are in the pricing table on this page.

How do I get a GPT-5.6 Luna API key?

Sign up at api.reapi.ai and create a key under API Keys, then pass it as a Bearer token in the Authorization header. Chat keys and their token balance are separate from the reapi.ai image and video gateway.

GPT-5.6 Luna — 1.05M Context at the Volume Tier

GPT-5.6 Luna is OpenAI's cost-sensitive tier of the GPT-5.6 family, and the interesting part is what it does not give up: GPT-5.6 Luna carries the same 1,050,000-token context window, the same 128,000-token maximum output and the same six-rung reasoning dial as its frontier sibling. What changes is the price and the depth of thinking you would sensibly ask of it. On reAPI, GPT-5.6 Luna bills 20% below OpenAI's published rate on both input and output.

GPT-5.6 Lunamodelgpt-5.6-luna

GPT-5.6 Luna playground

Chat models run on the api.reapi.ai gateway, which has its own console and its own key. Open the console to run GPT-5.6 Luna through the OpenAI-compatible Chat Completions surface.

Open console Token rates

What is the GPT-5.6 Luna API?

GPT-5.6 Luna is OpenAI's cost-sensitive tier of the GPT-5.6 family, served on reAPI through an OpenAI-compatible /v1/chat/completions endpoint. It gives up none of the family's limits — the same 1,050,000-token context window, the same 128,000-token maximum output and the same six reasoning-effort rungs from none to max, defaulting to medium — with text and image input and text output. Billing is per 1M tokens, with input and output priced separately; reasoning tokens count as output.

What you can build with this model

Real-world workflows and production use cases you can build and ship with this model.

GPT-5.6 Luna classifying a queue of documents with reasoning switched off

Classification and extraction, priced for the whole queue

This is the work OpenAI names for the lowest effort setting: fast information retrieval and classification, where reasoning adds latency and nothing else. Turn effort off, constrain the output with a schema, and GPT-5.6 Luna becomes a component rather than an oracle — the same prompt over a million rows, at a rate where the arithmetic still works. The part worth noticing is that dropping to this tier does not shrink your inputs: GPT-5.6 Luna reads the same 1,050,000-token window as the frontier tier, so a long document does not force you back up the price ladder just to fit.

Read the API docs

GPT-5.6 Luna triaging requests and escalating only the hard ones

The cheap first pass in front of an expensive model

Most production traffic is not hard. A two-stage route built on GPT-5.6 Luna — it reads everything and decides what is routine, a dearer model sees only what is not — is the single most effective cost lever available on a chat workload, and it works precisely because the cheap stage here is not a weak model but a smaller tier of the same family with the same limits and the same interface. Same key, same endpoint, one model string apart: escalation is a field in your request, not a second integration.

GPT-5.6 Luna answering a latency-critical request with no reasoning pass

Latency-critical paths where thinking is the wrong tool

Voice, autocomplete, live moderation, anything a human is waiting on: OpenAI's own advice for these is to try low effort first and move to none if required. GPT-5.6 Luna is the tier where that advice is cheapest to follow, and where the speed rating is Fast to begin with. Keep the prompt short, keep the output bounded, and treat effort as the dial you turn before you go looking for a different provider — the family's fast path and its deep path are the same API.

Pricing

Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.

First test cost

$0.8800 credits

One sample at the cheapest tier (1M tokens).

Testing budget guide

Add credits

$10

≈ 12 tests

$50

≈ 62 tests

$100

≈ 125 tests

Model	Category	Price
Token pricing	Input	$0.8 1M tokens
Token pricing	Output	$4.8 1M tokens

Why reAPI

20% below the published rate

OpenAI publishes a per-million-token rate for GPT-5.6 Luna input and output. On reAPI, GPT-5.6 Luna bills 20% below both — the same discount on each dimension, so your existing cost model scales down by one figure. No subscription underneath it, no minimum commitment, pay-as-you-go in USD against your gateway balance. The exact numbers are in the pricing table on this page.

OpenAI-compatible drop-in

GPT-5.6 Luna is served from a standard `/v1/chat/completions` endpoint. If your code already speaks that shape, adopting GPT-5.6 Luna is a base URL, a key and a model string — the same SDKs work once the base URL points at the gateway. No SDK swap, no bespoke client.

One key across the whole ladder

The same api.reapi.ai key that calls GPT-5.6 Luna calls its larger siblings, the Claude and Gemini families, and the open-weight tiers. That is what makes tiered routing practical: the escalation target is a model string, not a second vendor relationship with its own invoice and credentials.

GPT-5.6 Luna vs GPT-5.6 Terra

This is the comparison that actually decides a route, and it is unusually clean because GPT-5.6 Luna and GPT-5.6 Terra share almost everything: same context window, same output ceiling, same effort ladder, same tools, same knowledge cutoff, same endpoint. What differs is the reasoning headroom OpenAI rates each for, the lineage they descend from, and the price. Read the rows below as: does this route need judgement, or does it need throughput?

Capability

GPT-5.6 Luna on reAPI

GPT-5.6 Terra

OpenAI's positioning

Optimised for cost-sensitive, high-volume workloads

Balances intelligence and cost

Reasoning rating

High — the family's entry rating

Higher — one step up

Earlier-generation lineage

Corresponds to the nano tier of earlier GPT-5 families

Corresponds to the mini tier

Rate on reAPI

The lowest of the three tiers — see the pricing table above

Higher than Luna, lower than the frontier tier

Context and output limits

1,050,000-token context, 128,000-token maximum output

Identical — the limits do not change across tiers

Effort ladder and tools

All six rungs, functions, web search, file search, computer use

Identical — the difference is which rungs are worth paying for

Comparison reflects behavior documented on OpenAI's own model pages at the time of writing. Rates are described as relationships rather than amounts; the live numbers for GPT-5.6 Luna are in the pricing table above.

Ship GPT-5.6 Luna in three steps

step 01
Create an account and key on api.reapi.ai
Sign up at api.reapi.ai, open the console and generate an API key. That one key reaches GPT-5.6 Luna and every other model on the gateway, billed from a single balance.
Open
step 02
Point your client at the gateway
Set your base URL to api.reapi.ai and your key to the one you just created. Any client that already speaks OpenAI Chat Completions works unchanged — no SDK swap is needed to reach GPT-5.6 Luna.
Open
step 03
Send `gpt-5.6-luna` as the model
Set the model field to `gpt-5.6-luna` — with the dots — and post your messages. On this tier, set reasoning effort deliberately: the family default is medium, and on high-volume routes none or low is usually what you actually want.
Open

docs/api/gpt-5-6-luna

API reference

Drop-in code and the full parameter table.

View full reference

curl https://api.reapi.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.6-luna",
    "group": "default",
    "messages": [
      { "role": "user", "content": "Hello" }
    ],
    "reasoning_effort": "low",
    "stream": true
  }'

Parameter validation failed

Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.

Authentication or balance issue

401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.

Content or material rejected

Reference material and outputs pass automated moderation. Rejected tasks fail with a clear error and a full refund.

Task failed or timed out

A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.

Frequently asked questions

Common questions about this model.

GPT-5.6 Luna is billed pay-as-you-go in USD per token against your api.reapi.ai balance, with separate input and output rates and no subscription. Both sit 20% below OpenAI's published per-token rate. Current numbers are in the pricing table on this page.