
OpenAI
GPT-5.4
From $1.00 per 1M tokens
GPT-5.6 Terra is the tier OpenAI positions between intelligence and cost, and it lands exactly where the family's default reasoning effort sits. GPT-5.6 Terra reads a 1,050,000-token context window, returns up to 128,000 tokens, and takes all six effort rungs from off to max. OpenAI rates its reasoning Higher and its speed Fast. On reAPI, GPT-5.6 Terra bills 20% below OpenAI's published rate on both input and output.
gpt-5.6-terraChat 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 Terra through the OpenAI-compatible Chat Completions surface.
Real-world workflows and production use cases you can build and ship with this model.

Read OpenAI's own description of medium effort and it names agentic coding first, then research, spreadsheets and slides, and delegating long-horizon work. That list is the job description of a workhorse, and GPT-5.6 Terra is the tier priced for running it all day rather than for the occasional hard call. The family's limits do not change here: GPT-5.6 Terra still reads 1,050,000 tokens and still returns up to 128,000, so an agent working from a whole repository does not need a different tier just to see its inputs.
Read the API docs
Spreadsheets with working formulas, slide decks that follow a convention, runbooks with owners and rollback points: OpenAI puts this class of work under the same default effort setting, and it is where a 128,000-token output ceiling stops being a specification and starts being the reason the deliverable arrives whole. Pair GPT-5.6 Terra with a JSON schema when the artefact is consumed by another system, and with plain prose when a person will read it — the model is the same either way, only the contract changes.

Because GPT-5.6 Terra starts at the family default, it is the natural place to run the experiment that actually saves money: take a real route, try low, try high, and keep whichever your evals prefer. OpenAI's guidance for low is planning and multi-step decisions at speed, and for high it is complex debugging and deep planning — both reachable on this tier without changing models. When a route turns out to want max effort every time, that is the signal to move it up to the frontier sibling rather than pay for depth this tier was not priced for.
Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.
| Category | Unit | Price |
|---|---|---|
| Token pricing | ||
| Input | 1M tokens | $2 |
| Output | 1M tokens | $12 |
OpenAI publishes a per-million-token rate for GPT-5.6 Terra input and output. On reAPI, GPT-5.6 Terra bills 20% below both — the same discount on each dimension, so an existing estimate 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.
GPT-5.6 Terra is served from a standard `/v1/chat/completions` endpoint. If your code already speaks that shape, adopting GPT-5.6 Terra 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.
The same api.reapi.ai key that calls GPT-5.6 Terra calls the tier below it, the frontier tier above it, the Claude and Gemini families, and the open-weight models. Tiering a workload up or down is a model string rather than a second integration, a second invoice and a second set of credentials.
Same family, adjacent tiers, and an unusually honest comparison because nearly every specification is shared: identical context window, identical output ceiling, identical effort ladder, identical tools, identical knowledge cutoff, identical endpoint. What differs is the reasoning rating OpenAI gives each, the earlier-generation tier they descend from, and the rate. The question the rows below answer is whether your route needs the top of the ladder often enough to pay for it.
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 Terra are in the pricing table above.
Sign up at api.reapi.ai, open the console and generate an API key. That one key reaches GPT-5.6 Terra and every other model on the gateway, billed from a single balance.
OpenSet 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 Terra.
OpenSet the model field to `gpt-5.6-terra` — with the dots — and post your messages. You can leave reasoning effort alone to start with, since the family default is medium and that is this tier's intended setting; turn on streaming for anything a person is watching.
OpenCommon questions about this model.
Try it in the playground or grab an API key to integrate now.