runlanepreview
Open console

SHARED INFERENCE FOR DEVELOPERS

Give your agent
a place to run.

Open models for the way you build. Connect your coding tools, bring the context, and start with one API key.

Create an account Read the quickstart

Free launch trial · Shared capacity · No payment required

Checking model
QUICKSTART
from openai import OpenAI
import os

client = OpenAI(
    base_url="http://localhost:3080/v1",
    api_key=os.environ["RUNLANE_API_KEY"],
)

stream = client.chat.completions.create(
    model="YOUR_MODEL_ID",
    messages=[{"role": "user",
               "content": "Explain this code."}],
    max_tokens=2048,
    stream=True,
)
for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "",
              end="", flush=True)
Use your existing client. Change the endpoint.
MODEL ACCESSOpen models See the live catalog
CONTEXT WINDOW262,144 input + output tokens
INTERFACESOpenAI + Anthropic Text & function tools
Explore models and the roadmap

FROM ACCOUNT TO FIRST RESPONSE

A short path
to a running agent.

No model downloads or GPU setup on your side. Your requests go through one consistent endpoint.

Connection guide
  1. 01

    Create your workspace

    Create your account, verify your email and open the console. Join the launch trial without payment.

  2. 02

    Issue a key for your tool

    Name it, copy it once and keep it in an environment variable. Revoke keys independently.

  3. 03

    Run, inspect, repeat

    Stream responses and function calls. Inspect request status, token usage and queue time in your console.

Room for context.
Rules you can read.

The shared plan has no daily or monthly request quota. Each workspace runs one request at a time, up to six starts per minute.

Prompts above 32K input tokens are limited to six per hour. Requests may queue for up to 45 seconds. Output is capped at 32K tokens, within the 262K total window.

See every limit

START WITH YOUR OWN WORKFLOW

Put the endpoint to work.

Get your API key