The direct lane from your agents to every model.
One OpenAI-compatible endpoint in front of the major model families. Billed by the token, settled on Robinhood Chain. No sign-up, no plan, no card on file.
Usage counters start with the first metered call at launch.
One endpoint.
Priced per token.
Inferlane has no tiers and no credit bundles. A request costs the tokens it consumed at the posted rate, paid on Robinhood Chain. An agent can pay for a single call without ever opening an account.
answered from the lane cache
Pay per call
Call without a key and the lane answers with a price. Sign it, resend, receive the reply. Settled on Robinhood Chain with no account behind it.
Prepaid keys
Fund once and receive an il_sk_ key for any OpenAI SDK. Each call draws down the exact token cost.
Lane cache
A repeated deterministic request is served from cache at half the charge. Agents ask the same thing more often than you would guess.
Unspent budget returns
A paid call signs a ceiling, not a price. What the call did not use goes back to the wallet that paid, and closing a key returns its balance.
Newest on the lane
Indicative rates in USD per 1M tokens, taken from public list prices. Inferlane posts its own rates at launch.
glm-5.3-prime
Zhipu AI · 1M context
qwen3.8-max-prime
Qwen · 1M context
gpt-6-luna-pro
OpenAI · 1.1M context
gpt-6-luna
OpenAI · 1.1M context
$INFERLANE launches on Robinhood Chain.
The token of the Inferlane network. The address below is the only contract we will ever publish. Treat any other as someone else's.
0xxxxxxxxxxxxxxxxxxxxxxxxxxxxxYour SDK already
knows the way.
A wallet is the account
No sign-up form. The wallet that funds a key owns it, and the key is all your code carries.
Fund a key once
Mint a prepaid il_sk_ key from the dashboard. It is displayed a single time, so store it.
Change the base URL
Point your existing OpenAI client at the lane. Or drop the key entirely and let the agent pay each call over x402.
from openai import OpenAI client = OpenAI( base_url="https://inferlane.xyz/v1", api_key="il_sk_...", ) answer = client.chat.completions.create( model="meta-llama/llama-3.3-70b-instruct", messages=[{"role": "user", "content": "What is x402?"}], ) print(answer.choices[0].message.content)
Your agents know where to go. Give them the lane.
Billing is tokens used × posted rate, settled on Robinhood Chain. No per-request fee, nothing recurring, nothing tucked away.