GPT-5.5 API — Access via an AI API Gateway

Call OpenAI's gpt-5.5 through byesu — an AI API gateway with OpenAI-compatible /v1/chat/completions and /v1/responses endpoints. One key covers GPT-5.5, Claude, Grok, Gemini and more, billed pay-as-you-go with no subscription. byesu serves the official model with no quality degradation — no request rewriting, no caching of your business logic — and the Codex CLI plugs in directly over the Responses API.

On byesu, gpt-5.5 belongs to the OpenAI GPT group, and billing is per token × group multiplier:

Input 2.5x
Output 6x
Cache hit 0.1x (context hits automatically cut cost)

The multiplier applies to the model's base unit price. The actual per-1K-token billing rate is whatever the console token page and model list show in real time (the console displays the live billed price alongside the official list price for comparison). In practice: input takes the smallest share of the bill, output takes the largest, and cache hits cost almost nothing — the more long-context reuse, the better the value.

Pick the right group when creating a token

Before calling gpt-5.5, go to Console → Tokens and create a token in the OpenAI GPT group. Using the wrong group triggers a "no available channel" error. See Choosing a group for details.

In one line: what it is and how billing works

gpt-5.5 = OpenAI's latest flagship model, strong across both general reasoning and coding. On byesu you call it with your own token, pay-as-you-go — you're charged only for what you use, with no monthly fee and no minimum spend. Only images/video need the separate media group; text models like gpt-5.5 just use a token from the OpenAI GPT group.

How to call it

Base URL (base_url): https://byesu.com/v1Auth header: Authorization: Bearer sk-your-tokenEndpoints: OpenAI-compatible /v1/chat/completions (chat) and /v1/responses (Responses API — Codex uses this one).

First, create a token in the OpenAI GPT group (shaped like sk-xxxx), then replace sk-your-token in the examples below with it.

curl (/v1/chat/completions)

bash

curl https://byesu.com/v1/chat/completions \
  -H "Authorization: Bearer sk-YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.5",
    "messages": [
      {"role": "system", "content": "You are a concise assistant."},
      {"role": "user", "content": "Explain what an API gateway does in one sentence."}
    ]
  }'

The response is the standard OpenAI structure, with the body in choices[0].message.content:

json

{ "choices": [ { "message": { "role": "assistant", "content": "..." } } ] }

Python (openai SDK — zero changes, just swap base_url)

The official openai library works out of the box; just point base_url at byesu:

python

from openai import OpenAI

client = OpenAI(
    api_key="sk-YOUR_TOKEN",
    base_url="https://byesu.com/v1",
)

resp = client.chat.completions.create(
    model="gpt-5.5",
    messages=[
        {"role": "system", "content": "You are a concise assistant."},
        {"role": "user", "content": "Explain what an API gateway does in one sentence."},
    ],
)
print(resp.choices[0].message.content)

Streaming output (SSE)

The chat endpoint supports streaming — add "stream": true (curl) or stream=True (Python SDK) to get token-by-token output; all other parameters stay the same.

Responses API / Codex (/v1/responses)

For clients that speak the OpenAI Responses protocol (such as the Codex CLI), point the endpoint at https://byesu.com/v1 with wire_api = responses; the endpoint is /v1/responses. To call it directly with curl:

bash

curl https://byesu.com/v1/responses \
  -H "Authorization: Bearer sk-YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.5",
    "input": "Write a Python function that checks whether a string is a palindrome."
  }'

Key parameters

Parameter	Required	Description
`model`	✅	Always set to `gpt-5.5`; it must match the console model name exactly (case and hyphens included), or you'll get a "model does not exist" error
`messages`	✅ (chat)	The conversation array; `role` is one of `system`/`user`/`assistant`. For `/v1/responses`, use `input` instead
`stream`		`true` enables token-by-token SSE streaming; defaults to `false`
`temperature`		Sampling temperature — lower is more deterministic, higher is more varied; model default if unset
`max_tokens`		Max tokens generated per reply (affects output billing, charged at the 6x multiplier)
`reasoning_effort`		Reasoning intensity: `minimal`/`low`/`medium`/`high` (on the Codex side it's `model_reasoning_effort`); higher is slower and pricier but higher quality
`response_format`		Structured output, e.g. `{"type": "json_object"}` to force JSON

Billing note: input 2.5x, output 6x, cache hit 0.1x. Tokens that hit the context cache are billed at 0.1x — the more you reuse across long conversations and long prompts, the more you save.

Why use byesu to access gpt-5.5

Official routing, no quality degradation: Requests reach OpenAI's official service (not a reverse-engineered, mirrored, or cracked stand-in), so the model's capabilities match the official service — no sneaky downgrades.
Transparent pricing: Pay-as-you-go with public multipliers (input 2.5x / output 6x / cache 0.1x); the console shows the live billed price in real time alongside the official list price. You pay only for what you use — no monthly fee, no minimum spend.
Cache savings: Context-cache hits are billed at 0.1x, cutting costs significantly for long context, multi-turn conversations, and repeated prompts.
OpenAI-compatible, zero migration: Full compatibility with /v1/chat/completions and /v1/responses; with the official openai SDK you change a single line — base_url — and existing code needs no rewrite.
Direct hookup for Codex / Claude Code: Point the Codex CLI at https://byesu.com/v1 (wire_api=responses) and it works; the same console also connects Claude Code, Cherry Studio, and 17 clients in total — one account, reused across many endpoints.
Flexible top-ups: Top up with USDT, Alipay or WeChat Pay — pay only for tokens you use.

Quick Start (3-step setup) — get a key, set the URL, pick a model
Codex CLI integration — full config for running gpt-5.5 over the Responses API
Choosing a group — how to create the OpenAI GPT group
Common errors & fixes — troubleshooting 401 / insufficient balance / no available channel
Register / create a token in the console — create an OpenAI GPT group token, top up, and view real-time billing rates

FAQ

How do I get gpt-5.5 API access?

Sign up at the byesu console, create a token in the OpenAI GPT group, and call gpt-5.5 through the OpenAI-compatible /v1/chat/completions or /v1/responses endpoint. Set the base URL to https://byesu.com/v1 and authenticate with Authorization: Bearer sk-xxxx. One key also covers Claude, Grok, Gemini and more, billing is pay-as-you-go with no subscription, and you can top up with USDT, Alipay or WeChat Pay.

How much does the gpt-5.5 API cost?

Pay-as-you-go, with no monthly fee and no minimum spend. On byesu, gpt-5.5's group multipliers are: input 2.5x, output 6x, and cache hit 0.1x (the multiplier applies to the model's base unit price). The live billed price per 1K tokens is shown in real time on the console token page and model list, next to the official list price for comparison. Long-context tokens that hit the cache are billed at 0.1x, which noticeably lowers cost.

How do I use gpt-5.5 in Codex?

Point Codex at byesu: in config.toml set model = "gpt-5.5", base_url = "https://byesu.com/v1", wire_api = "responses", and add disable_response_storage = true; in auth.json set OPENAI_API_KEY to your byesu token. See the Codex CLI integration page in the docs for details.

How do I use gpt-5.5 in Claude Code / other clients?

For any OpenAI-compatible client (Cherry Studio, Codex, etc.), set base_url to https://byesu.com/v1, model to gpt-5.5, and the key to your token. Claude Code is mainly for Claude-family models (use base_url https://byesu.com); to use gpt-5.5 we recommend Codex or any OpenAI-compatible client. byesu supports 17 clients — one account reused across many endpoints.

Does byesu serve the official gpt-5.5 model? Will quality be degraded?

Yes. byesu uses official routing — not reverse-engineered, mirrored, or cracked — so the model's capabilities match OpenAI's official service with no quality degradation. The interface is the standard OpenAI-compatible format, and migrating with the official openai SDK takes just a one-line base_url change — no rewriting of existing code.

Is calling gpt-5.5 through byesu the same as calling the official API?

Yes. byesu provides OpenAI-compatible /v1/chat/completions and /v1/responses endpoints, and parameters like messages, stream, temperature, max_tokens, and response_format match the official API. Just swap base_url to https://byesu.com/v1 and the key to your byesu token — the rest of your code stays unchanged.

GPT-5.5 API — Access via an AI API Gateway ​

In one line: what it is and how billing works ​

How to call it ​

curl (/v1/chat/completions) ​

Python (openai SDK — zero changes, just swap base_url) ​

Streaming output (SSE) ​

Responses API / Codex (/v1/responses) ​

Key parameters ​

Why use byesu to access gpt-5.5 ​

Related links ​

FAQ ​

How do I get gpt-5.5 API access? ​

How much does the gpt-5.5 API cost? ​

How do I use gpt-5.5 in Codex? ​

How do I use gpt-5.5 in Claude Code / other clients? ​

Does byesu serve the official gpt-5.5 model? Will quality be degraded? ​

Is calling gpt-5.5 through byesu the same as calling the official API? ​