GPT-6 Luna is OpenAI’s most efficient model for focused, high-volume tasks. GPT-6 Luna reads a 1,050,000-token context window, returns up to 128,000 tokens, takes text and images, and lets you set reasoning effort from none to max. On reAPI, GPT-6 Luna is billed per token below OpenAI’s published rate.
Start with the prompt in the composer or choose an example.
Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.
One 1K-token prompt with a 1K-token answer at the cheapest tier.
| Category | Standard · prompts up to 272K tokens | Long context · prompts over 272K tokens |
|---|---|---|
| Input | $0.080 Price · 1M tokens | $0.160 Price · 1M tokens |
| Output | $0.400 Price · 1M tokens | $0.600 Price · 1M tokens |
| Cache read | $0.008 Price · 1M tokens | $0.160 Price · 1M tokens |
| Cache write | $0.100 Price · 1M tokens | $0.200 Price · 1M tokens |
GPT-6 Luna is the most efficient model in OpenAI’s GPT-6 line, built for focused, high-volume tasks. GPT-6 Luna accepts text and images and produces text. GPT-6 Luna suits work that repeats thousands of times a day — sorting tickets, pulling fields out of documents, tagging content, routing requests — where each call has a clear, checkable answer and the per-token rate matters. On reAPI, GPT-6 Luna is served through the OpenAI-compatible Chat Completions endpoint and billed per token.
Real-world workflows and production use cases you can build and ship with this model.
Send GPT-6 Luna each incoming support ticket, email or form entry with a fixed list of labels and ask for one label plus a short reason. Keep the label set small and defined in the system message so GPT-6 Luna results stay consistent across millions of calls. Run GPT-6 Luna at a low reasoning effort for plain sorting, spot-check a sample against human labels, and route the uncertain cases to a person or a larger model.
Read the API docs
Give GPT-6 Luna invoices, receipts, contracts or screenshots and ask for the exact fields your system needs. Structured outputs let you pin the reply to a JSON schema, and image input means a scanned page or a photographed receipt can go in directly. GPT-6 Luna returns the shape you asked for; validate the values against your own records before they reach a ledger or a customer.
Put GPT-6 Luna in front of a larger workflow to decide what each request needs: which team, which tool, which model. A short prompt, a list of options and function calling with reasoning_effort set to none keep each GPT-6 Luna decision quick. Hand the hard cases on to GPT-6 Sol and let GPT-6 Luna absorb the bulk of the traffic.
OpenAI positions GPT-6 Luna as its most efficient GPT-6 model, and its published per-token rate sits far below GPT-6 Sol’s. reAPI bills GPT-6 Luna below that published rate again. For workloads measured in millions of calls — tagging, extraction, triage — the per-token rate decides the bill, so GPT-6 Luna is the first model to test.
Efficiency does not mean a smaller window. OpenAI documents the same 1,050,000-token context window, 922,000 maximum input tokens and 128,000 maximum output tokens for GPT-6 Luna as for GPT-6 Sol. GPT-6 Luna can read a long document in one call when a task needs it, then answer in a few tokens.
On reAPI, GPT-6 Luna takes all six reasoning effort levels, from none through max, with medium as the default. Plain GPT-6 Luna sorting and extraction jobs often need none or low; a harder judgement call can take more. Reasoning tokens bill as output, so measure which setting holds your accuracy target and use the lowest one that does.
GPT-6 Luna and GPT-6 Sol share the same context and output limits. The difference is the job: GPT-6 Luna is tuned for efficient, high-volume work, GPT-6 Sol for complex coding and agents.
Source: OpenAI model documentation, reviewed September 24, 2026. Official specifications do not establish a measured quality difference; test both on your own tasks.
Sign in to reAPI and create a key under API keys. The same key reaches GPT-6 Luna and every other model on the platform, billed from one balance.
OpenSet the base URL to https://reapi.ai/api/v1 and use the key you just created. Any client that already speaks OpenAI Chat Completions can call GPT-6 Luna without an SDK change.
OpenSet the model field to gpt-6-luna and post your messages. Pick the lowest reasoning_effort that meets your accuracy target; requests with tools work at every setting.
Opencurl https://reapi.ai/api/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-6-luna",
"messages": [
{ "role": "user", "content": "Hello" }
],
"reasoning_effort": "medium",
"max_completion_tokens": 16000,
"stream": true
}'Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.
401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.
Reference material and outputs pass automated moderation. Rejected tasks fail with a clear error and a full refund.
A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.
Common questions about this model.
Try it in the playground or grab an API key to integrate now.
