Passion8 API

Passion × Infinity

Driven by passion · Infinite creation

Connect AI · Power Creation

API RELAY / MODEL GATEWAY

Smarter relay,faster creation

Full-power AI, within reach

Use frontier AI models for a fraction of the official price.

Passion8 is an AI API relay for developers, creators, and AI tool users. Connect OpenAI-compatible APIs, Claude Code, Codex, Gemini, and CC-Switch through one stable base URL.

Start integratingRead docs
anthropic_sdk.py
# Keep the official SDK. Replace only the gateway.
from anthropic import Anthropic

client = Anthropic(
    api_key="sk-...",
    base_url="https://passion8.cc",
)

resp = client.messages.create(
    model="claude-opus-4-8",
    max_tokens=1024,
)
curl.sh
# Call Passion8 directly with curl.
curl https://passion8.cc/messages \
  -H "x-api-key: sk-..." \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "hello"}]
  }'
¥=$
official price, in RMB
<1s
TTFT target
7x24
stable relay
99.9%
SLA

QUICKSTART

Keep your tools. Replace the endpoint.

01

Point Base URL

Use one relay endpoint for OpenAI-compatible and Claude-style clients.

base_urlset
02

Swap the key

Keep SDK habits unchanged while Passion8 manages billing and routing.

api_keyverified
03

Start calling

Run Codex, Claude Code, scripts, and agents with predictable latency.

status200 ok

MODEL TABLE

Model routing without vague platform talk

Compare the line items that matter before routing production traffic: input, output, cache reads, and the client each model fits best.

Flagship codingAnthropic

claude-opus-4-8

Input · per 1M tokens

Official

$5

Passion8

¥5

Output · per 1M tokens

Official

$25

Passion8

¥25

Cache read

¥0.5

Cache write

¥6.3

Best clientClaude Code
Frontier reasoningOpenAI

gpt-5.5

Input · per 1M tokens

Official

$2

Passion8

¥2

Output · per 1M tokens

Official

$8

Passion8

¥8

Cache read

¥0.2

Cache write

¥2

Best clientOpenAI SDK
Long contextGoogle

gemini-2.5-pro

Input · per 1M tokens

Official

$1.25

Passion8

¥1.25

Output · per 1M tokens

Official

$10

Passion8

¥10

Cache read

¥0.31

Cache write

¥1.6

Best clientGemini clients
Fast & cheapOpenAI

gpt-5-mini

Input · per 1M tokens

Official

$0.25

Passion8

¥0.25

Output · per 1M tokens

Official

$2

Passion8

¥2

Cache read

¥0.03

Cache write

¥0.3

Best clientCodex CLI

RELAY SYSTEM

Built for people who ship with AI tools every day

Official model access

Full GPT-5, Claude, and Gemini line-up — original models, no quality downgrade, no wrappers.

Transparent pricing

Pay the official number in RMB instead of USD, billed per token with no hidden margin.

Global acceleration

Cloudflare edge network with multi-line BBR routing keeps latency low and calls reachable.

Secure key handling

Independent API keys, encrypted at rest, over full HTTPS transport — privacy first.

BASE URL CHECKLIST

Choose a client, copy the working example

No platform buzzwords — just what you can actually use: real tools, real base URLs, real configuration you can copy and run.

Anthropic-compatible clients keep their SDK calls. Replace the base URL and API key.

Python (Anthropic)
import os
from anthropic import Anthropic

client = Anthropic(
    api_key=os.environ["PASSION8_API_KEY"],
    base_url="https://passion8.cc",
)

resp = client.messages.create(
    model="claude-opus-4-8",
    max_tokens=1024,
    messages=[{"role": "user", "content": "hello"}],
)

print(resp.content[0].text)

FEEDBACK

Real feedback from frontline developers

Prices match official logic and reconciliation finally makes sense.
YYuki · AI PM
One base_url solved our OpenAI access problem and passed compliance.
HHan · Tech Lead
No Stripe, still got Claude running. We shipped a demo in 10 minutes.
AAJ · Indie Developer
Recharge lands instantly, and midnight batch runs no longer get stuck.
HHe · SaaS Founder
No Stripe, still got Claude running. We shipped a demo in 10 minutes.
AAJ · Indie Developer
Recharge lands instantly, and midnight batch runs no longer get stuck.
HHe · SaaS Founder
We changed one Anthropic SDK base_url and kept CI/CD untouched.
JJane · Frontend Engineer
Support answers fast, and real engineers help debug hard issues.
KKe · CTO

OUR NAME

Passion plus infinity

Passion8 reads as passion and the infinity of 8 — the belief that when access stops getting in the way, what people can build has no ceiling. We keep the path between your tools and the best models short, so the work stays on the ideas.

Passion × Infinity

Driven by passion · Infinite creation

01

Global

Built for developers and creators who need stable access.

02

Developer first

Configuration is short, visible, and easy to verify.

03

Creation focused

Lower friction means more ideas reach working software.

Frequently asked questions

Do I need to change my code?

No. Keep your existing SDK and request shape — point the base URL at Passion8 and swap in your key. Anthropic-style and OpenAI-compatible clients both work unchanged.

Which models and clients are supported?

GPT-5, Claude, Gemini and new flagship models, used from Claude Code, Codex, CC-Switch, the OpenAI SDK, and your own agents through one stable endpoint.

How is billing calculated?

Token-based, settled against official model pricing with no hidden margin. Usage, cache reads and per-request cost stay visible so reconciliation matches what you expect.

Is my key and traffic secure?

Each key is independent and revocable, all traffic is HTTPS, and usage boundaries are explicit. Keys never need to leave your environment beyond the request itself.