Claude · GPT · Gemini · image generation — one key

One API key for Claude, GPT and Gemini.
Pay only for what you use.

Text models from Anthropic, OpenAI and Google plus image generation, behind one OpenAI- and Anthropic-compatible endpoint. No subscription and no monthly fee: every request and what it cost shows up in your logs.

Get your API key Read the quickstart

OpenAI- and Anthropic-compatible API Every request itemized in your logs Public status page Sign up with just an email

Why developers use Byesu

One account instead of separate vendor accounts, with the balance, the charges and the health of every line in front of you.

One key, every major model

Claude, GPT, Gemini, DeepSeek, Kimi and GLM, plus image models. Switch models by changing one string in your code.

Transparent multipliers

Each model bills as a multiplier of the vendor's official list price, usually a small fraction of it. Cached input is billed at as little as 0.1×.

Nothing hidden

Every request is logged with model, tokens and charge. Availability of every line is on a public status page, measured on real traffic rather than synthetic pings.

Drop-in for your tools

Claude Code, Codex CLI, Cursor, Cline and any OpenAI or Anthropic SDK: change the base URL and the key, keep your code.

Models you can call today

A selection of the current lineup. The full list, with the multiplier for each model, is in the console.

Anthropic

claude-opus-5-5
claude-sonnet-5-5
claude-haiku-5-5
claude-fable-5-1

OpenAI

gpt-6.1-sol
gpt-6.1-terra
gpt-6-sol
gpt-5.6-sol

Google

gemini-4-argon
gemini-3.8-flash
gemini-3.6-flash
gemini-3.1-pro-low

deepseek-v4-pro
kimi-k3
glm-5.3

No subscription and no minimum monthly spend: you are charged only for the tokens you use. Prompt caching works the vendor's way, and cache reads on Claude are billed at 0.1× of input. If you set cache breakpoints badly, caching can cost more rather than less, so check your logs.

About lines. A model can be served by more than one line, and lines differ in multiplier and capacity. You choose a line when you create a key, or use auto and let the gateway pick. Each line's live availability is on the status page.

Images: five models, one price at any resolution

Text-to-image and image-to-image (up to 6 references) through the OpenAI-compatible /v1/images endpoint or Gemini's native generateContent. 1K, 2K and 4K cost the same.

OpenAIgpt-image-2$0.009/ image1K · 2K · 4K

OpenAIgpt-image-2.5$0.009/ image1K · 2K · 4K

Googlegemini-3.0-pro-image-ultra$0.009/ imageNano Banana Pro · 1K · 2K · 4K

Googlegemini-nano-banana-2.1-ultra$0.008/ image1K · 2K · 4K

Googlegemini-3.1-flash-image-ultra$0.007/ image1K · 2K · 4K

Same price for 1K / 2K / 4K Up to 6 reference images Generate in the browser, no code needed Failed generations are generally not billed

Prices as of November 6, 2026. Formats, sizes and request examples: image API docs

Get started in three steps

Sign up

All you need is an email. Registration takes a minute.

Top up your balance

Crypto (USDT): in the console under Wallet, from any wallet.
Alipay: also under Wallet.

Create a key and connect

Under Tokens, create an sk-… key, pick a line (or auto) and use the base URL from the examples below.

# Python · OpenAI SDK (works for GPT, Claude and Gemini)
from openai import OpenAI
client = OpenAI(base_url="https://byesu.com/v1", api_key="sk-…")
r = client.chat.completions.create(
    model="claude-sonnet-5-5",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(r.choices[0].message.content)

# Claude Code / Anthropic SDK (native Anthropic protocol)
export ANTHROPIC_BASE_URL=https://byesu.com
export ANTHROPIC_AUTH_TOKEN=sk-…
claude

# curl
curl https://byesu.com/v1/chat/completions \
  -H "Authorization: Bearer sk-…" -H "Content-Type: application/json" \
  -d '{"model":"gpt-6.1-sol","messages":[{"role":"user","content":"Hello!"}]}'

Works with the tools you already use

Any client that lets you change the base URL. Step-by-step guides are in the docs.

Claude Code

Codex CLI Cursor Cline / Roo Code Cherry Studio CC Switch OpenAI / Anthropic SDK All clients →

Talk to us

Line and model changes are announced in the Telegram channel first. Questions go to the group chat or the support bot, and we post engineering notes on X.

Telegram @byesuE news Support bot private questions X @amdmszOvO notes

Frequently asked questions

Do I need an OpenAI or Anthropic account?

No. You sign up with an email, top up your Byesu balance and call every model with one key. Separate vendor accounts and vendor billing are not needed.

Are these the real models?

Requests go to the vendors' original models. Your logs show the model, tokens and charge of every request, and the status page shows availability from real traffic.

How is usage billed?

Text models bill per token as a multiplier of the vendor's official list price; you can see the multiplier of each model in the console before you call it. Image models are a flat price per image. There is no subscription.

Does it work with Claude Code, Cursor and Codex CLI?

Yes. Claude Code needs two environment variables (see the example above); for other clients, see the client guides.

What happens if a line goes down?

Check the status page and switch your key to another line, or use auto, which routes around lines that are failing.

Can I get a refund?

Refund requests for unused balance are reviewed case by case through the support bot. Failed image generations are generally not billed; if you see a charge you don't expect, contact support. Promotional credit has no cash value.

Where is Byesu available?

Byesu is not offered in mainland China or in comprehensively sanctioned regions, and vendor regional restrictions apply. See supported regions.

Start small and check the logs yourself

Get your API key Read the docs