Build on llm.buzz

Keep the SDK you already use. Change the base URL and the key.

Two lines in either SDK

Streaming, tools and the rest work as they are. Model names are the ones on llm.buzz, such as moonshotai/kimi-k3.

OpenAI SDK

import os
from openai import OpenAI
 
client = OpenAI(
Removed:     base_url="https://api.openai.com/v1",
Added:     base_url="https://api.llm.buzz/api/v1",
Removed:     api_key=os.environ["OPENAI_API_KEY"],
Added:     api_key=os.environ["LLM_BUZZ_KEY"],
)

Anthropic SDK

import os
from anthropic import Anthropic
 
client = Anthropic(
Added:     base_url="https://api.llm.buzz/api",
Removed:     api_key=os.environ["ANTHROPIC_API_KEY"],
Added:     api_key=os.environ["LLM_BUZZ_KEY"],
)

Make a request

The same call in three languages, against a model live on llm.buzz now.

curl

Added: curl https://api.llm.buzz/api/v1/chat/completions \
  -H "Authorization: Bearer $LLM_BUZZ_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "moonshotai/kimi-k3",
       "messages": [{"role": "user", "content": "Hello"}]}'

Python

import os
from openai import OpenAI
 
client = OpenAI(
Added:     base_url="https://api.llm.buzz/api/v1",
    api_key=os.environ["LLM_BUZZ_KEY"],
)
reply = client.chat.completions.create(
    model="moonshotai/kimi-k3",
    messages=[{"role": "user", "content": "Hello"}],
)

TypeScript

import OpenAI from "openai";
 
const client = new OpenAI({
Added:   baseURL: "https://api.llm.buzz/api/v1",
  apiKey: process.env.LLM_BUZZ_KEY,
});
const reply = await client.chat.completions.create({
  model: "moonshotai/kimi-k3",
  messages: [{ role: "user", content: "Hello" }],
});

Base URL: https://api.llm.buzz/api/v1
Header: Authorization: Bearer $LLM_BUZZ_KEY
Endpoints: POST /chat/completions
POST /responses
POST /messages
POST /embeddings
GET /models

llm.buzz API OpenAI SDK Anthropic SDK OpenRouter routes llms.txt llms-full .txt For AI

Every way in.

The SDKs you know, the OpenRouter routes, and plain files an agent can read.

The public catalog

Read-only JSON with no key: the data these pages are built from. It is the site’s data, not a stable API, so programs should call the model API.

GET /siteThe catalog’s counts, the standard rate and the billing facts.

GET /modelsEvery public model with its prices, context, capabilities and providers.

GET /models/{name}One model with each provider’s prices, limits, data policy and performance.

GET /providersEvery public provider.

GET /rankingsThe most used models and apps of the last complete weeks.

curl

curl https://api.llm.buzz/public/v1/models

For programs, GET https://api.llm.buzz/api/v1/models lists the public models without a key too, in OpenRouter’s shape.

Building something on llm.buzz?Ask questions and share what you made in the Surogate Discord.

Join the Discord