Every major model,
one prepaid balance.
OpenAI-compatible API. Claude, GPT, Gemini, Llama, DeepSeek — one endpoint, one balance, one bill. Bulk-purchased capacity means you pay significantly less than list price.
no credit card · no subscription · no minimums
Change two lines.
Keep your code.
If your code already talks to OpenAI, Anthropic, or Google via any of the
standard SDKs, point base_url at llmrelay and
swap your API key. Nothing else changes.
from openai import OpenAI
client = OpenAI(
base_url="https://api.llmrelay.dev/v1",
api_key="llm_...",
)
resp = client.chat.completions.create(
model="claude-opus-4-7",
messages=[{"role": "user", "content": "Hello"}],
)Prepaid, not subscription
Top up when you need it. Your credit doesn't expire. No monthly bill, no auto-renewal, no surprise charges on usage spikes.
One endpoint, every model
Route Claude, GPT, Gemini, Llama, and open-source models through the same OpenAI-compatible endpoint. Switch model with a string, not a rewrite.
Significantly less than list
We buy at wholesale rates and pass most of the savings on. Transparent per-token pricing, live on the pricing page.
All the models. One balance.
Route to whichever model fits the task. Frontier for hard problems, open-source for cheap bulk work.
Swap in in 30 seconds.
Sign up, get $0.20 in credit, run a few real requests, decide if it's worth the swap.
Get your API key →