Docs

OpenAI-compatible endpoint. Drop-in replacement for any code that currently uses openai.chat.completions.create() or equivalent. Change base_url and api_key, keep the rest.

Quickstart (Python)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.solo-stack.xyz/v1",
    api_key="ssk-...",  # your unlimited-llm key
)

response = client.chat.completions.create(
    model="llama-3.3-70b",  # or qwen-2.5-72b, gpt-4o-mini (Indie+), claude-haiku (Indie+)
    messages=[{"role": "user", "content": "ping"}],
)
print(response.choices[0].message.content)

Quickstart (Node)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.solo-stack.xyz/v1",
  apiKey: process.env.UNLIMITED_LLM_KEY,
});

const res = await client.chat.completions.create({
  model: "llama-3.3-70b",
  messages: [{ role: "user", content: "ping" }],
});

Available models

• llama-3.3-70b, self-hosted, unlimited on every tier
• qwen-2.5-72b, self-hosted, unlimited on Indie and Shop
• gpt-4o-mini, OpenAI fallback, capped per tier (Indie+)
• claude-haiku, Anthropic fallback, capped per tier (Indie+)
• gpt-4o, claude-sonnet, gemini-pro, fallback, capped (Shop)

Fallback behavior

Requests are routed to self-hosted GPU first. If the self-hosted model is saturated (rate limit exceeded for the tier), the request falls back to the matching frontier model on your tier. Fallback usage counts against a soft cap per tier, but does not result in per-token billing.

Rate limits

• Starter: 5 req/sec, self-host only
• Indie: 20 req/sec, self-host + capped frontier fallback
• Shop: 100 req/sec, self-host + all frontier fallback

Status

Live status will be at status.solo-stack.xyz. For now, the service is in early-access waitlist.