Copy a shared key below and start testing in seconds — no account, no card. Or create an account and get $20 of credit on a permanent key that reaches Claude Opus 5.
Or scroll down and copy a shared key — no account needed.
No signup required
These short-lived shared keys refresh automatically. Each has its own remaining credit, request limit and expiry, so try another key if one has already been used up.
Loading current shared keys…
OpenAI-compatible project · 3 highlighted model families · 90+ catalog routes · public X updates
Model Coverage
Use this catalog as a developer map for model families, providers and route coverage. The public project stays focused on access, examples and update signals.
Live model catalog is loaded from the production routing inventory.
| Model | Provider | Cloud | Status |
|---|
No models match this filter.
Loading live model catalog...
Coverage only. For keys and examples, read the docs or follow @getkeyway for availability notes.
Why follow
The landing page should send builders to docs and X because that is where the useful setup and availability updates happen.
Start from the authentication docs so test keys, base URLs and SDK examples stay in one controlled path.
@getkeyway is the fastest channel for testing windows, model alerts, uptime notes and short build updates.
Use the same OpenAI SDK pattern in Python, Node, cURL and common tools. Copy a key, swap the base URL and test.
Follow changes for GPT, Claude, Gemini, DeepSeek, Kimi, embeddings, image and audio routes without reading a storefront.
Verify whether a key still works before wiring it into your app, then jump into the docs when you need integration details.
Docs, the model catalog and the playground keep the setup path inspectable without depending on an external repository.
Quick Start
Replace base_url. Keep the rest of your code identical.
from openai import OpenAI
client = OpenAI(
base_url="https://aiapiv2.pekpik.com/v1",
api_key="sk-...",
)
resp = client.chat.completions.create(
model="claude-opus-4-7", # or gpt-5.4, deepseek-chat, gemini-2.5-pro
messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)Explore the build
These pages help developers understand model coverage, OpenAI-compatible migration and routing behaviour before wiring requests into an app.
Developer pages for understanding which model families the project can target.
Provider-level notes for developers mapping OpenAI-compatible requests to model families.
Integration, routing and reliability topics for builders using OpenAI-compatible clients.
Comparison pages for developers evaluating model access patterns and routing tradeoffs.
FAQ
Start with the authentication docs and use the OpenAI-compatible base URL shown there. Follow @getkeyway on X for fresh testing windows and availability notes.
No. Requests pass through without content logging. We retain only aggregate routing metrics such as model, latency and health signals.
@getkeyway on X is the fastest public update channel for testing windows, route notes and model availability.
Public testing access is temporary, so limits vary by route and usage. If one key is exhausted, check the docs or @getkeyway on X for current availability.
Treat public keys as a testing and learning resource. For production workloads, use your own account controls and stable operational limits.
The primary interface is OpenAI-compatible /v1/chat/completions. Provider-native endpoints can be evaluated when a workflow needs them.
Each model can be backed by a pool of upstream keys with continuous health checks. Bad keys are quarantined and traffic shifts to healthy ones automatically.
Follow the build
Start from the docs for key setup and examples. X is where @getkeyway posts availability notes, model alerts and fresh testing windows first.