GPT-5.5 is OpenAI's newest reasoning model — a drop-in /v1/chat/completions endpoint with 1M context, 128K max output, advanced reasoning with adjustable effort, and Tool Search for large agent workflows. Pay-as-you-go pricing in USD.
gpt-5.5Chat models run on the api.reapi.ai gateway, which has its own console and its own key. Open the console to run GPT-5.5 through the OpenAI-compatible Chat Completions surface.
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.
GPT-5.5 is OpenAI's frontier reasoning model, served on reAPI as a drop-in OpenAI-compatible /v1/chat/completions endpoint. It reads a 1M-token context window, returns up to 128K tokens per call, and adds advanced reasoning with adjustable effort plus Tool Search, which lets agents pull the right tools from a large registry instead of carrying every definition in every prompt. Billing is per 1M tokens, with input and output priced separately.
Real-world workflows and production use cases you can build and ship with this model.
Build research agents, decision-support systems, and long-running planners on the GPT-5.5 API. Advanced reasoning with adjustable effort (`none` to `xhigh`) lets you trade latency for depth on a per-call basis — production teams can dial reasoning down for cheap turns and up for the hard ones.
Read the API docs
Process entire repositories, long technical documents, or large research sets in a single request. The GPT-5.5 API's 1M token context window makes whole-codebase refactoring, audit, and generation practical from a single /v1/chat/completions call.
Wire GPT-5.5 into workflows that need many tools, internal APIs, or MCP connectors. Tool Search lets the model pull only the tools it needs on demand instead of dumping every definition into every prompt — agent quality improves and token waste drops.
The GPT-5.5 API speaks OpenAI Chat Completions verbatim. For most teams, moving an existing OpenAI integration is a base URL, an API key, and a model-string change — not a platform rewrite. The same `messages` array, the same `stream`, `temperature`, `top_p`, `frequency_penalty`, and `presence_penalty` parameters, the same SSE wire format.
GPT-5.5 is billed pay-as-you-go in USD — current per-1M-token rates are listed in the pricing card on this page so you can estimate production cost before routing real workloads. Top up once, your balance is drawn down per request, no monthly minimum.
A single api.reapi.ai key unlocks GPT-5.5 alongside Claude Opus 4.7, Gemini 3.1 Pro, and every other frontier chat model on the platform. That makes vendor comparison, fallback routing, and per-call cost optimisation a configuration change instead of an integration project.
Sign up at api.reapi.ai, open the console, generate an API key under API Keys, and top up tokens under Top Up. The chat workspace is separate from the reapi.ai image/video gateway — keys do not cross over.
OpenPOST https://api.reapi.ai/v1/chat/completions with `model: "gpt-5.5"`, your `messages` array, and any optional parameters. The endpoint uses the standard chat-completions format, including streamed responses.
OpenOpen the GPT-5.5 docs for the full request schema, the >272K long-context billing rule, error envelope, and language-specific code samples (cURL, Python, Node.js, Go).
Opencurl https://api.reapi.ai/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5.5",
"group": "default",
"messages": [
{ "role": "user", "content": "Hello" }
],
"stream": true,
"temperature": 0.7,
"top_p": 1,
"frequency_penalty": 0,
"presence_penalty": 0
}'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.
