Install
Pick how you want Plynf. Ranked by friction.
Most people are one copy or one click away. Pick what you already use below — or scroll for every install path, ranked by friction.
Plynf Cloud
No Docker, no servers. Get a hosted endpoint and an API key in one click — then connect your stack below. Free to start.
Start on Plynf CloudSelf-host
One command spins up the full stack locally in ~60s.
curl -fsSL https://plynf.com/install.sh | sh OpenAI · Python
AI SDKSwap the base URL. Every existing call is shaped + routed.
from openai import OpenAI
client = OpenAI(base_url="https://app.plynf.com/v1", api_key="plynf_sk_live_…")
# …your existing calls are unchanged. OpenAI · Node
AI SDKOne line in your client config. Works with the official SDK.
import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://app.plynf.com/v1", apiKey: "plynf_sk_live_…" }); Anthropic · Python
AI SDKPoint the Anthropic SDK at Plynf's native /v1/messages door.
from anthropic import Anthropic
client = Anthropic(base_url="https://app.plynf.com", api_key="plynf_sk_live_…") Anthropic · TS
AI SDKbaseURL on the Anthropic TS SDK — messages + token counting flow through.
import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic({ baseURL: "https://app.plynf.com", apiKey: "plynf_sk_live_…" }); Google Gemini
AI SDKSet the genai client's base URL — :generateContent routes via Plynf.
from google import genai
client = genai.Client(api_key="plynf_sk_live_…",
http_options={"base_url": "https://app.plynf.com"}) Mistral
AI SDKserver_url on the Mistral SDK points at Plynf.
from mistralai import Mistral
client = Mistral(api_key="plynf_sk_live_…", server_url="https://app.plynf.com/v1") Cohere
AI SDKbase_url on the Cohere v2 client — /v2/chat routes through Plynf.
import cohere
co = cohere.ClientV2(api_key="plynf_sk_live_…", base_url="https://app.plynf.com") Azure OpenAI
AI SDKPoint AzureOpenAI's endpoint at Plynf's Azure-shaped door.
from openai import AzureOpenAI
client = AzureOpenAI(azure_endpoint="https://app.plynf.com",
api_key="plynf_sk_live_…", api_version="2024-06-01") Vercel AI SDK
AI SDKCreate an OpenAI provider pointed at Plynf, use it anywhere.
import { createOpenAI } from "@ai-sdk/openai";
export const plynf = createOpenAI({ baseURL: "https://app.plynf.com/v1", apiKey: "plynf_sk_live_…" }); LangChain
FrameworkSet base_url on ChatOpenAI — chains, agents and tools all flow through.
from langchain_openai import ChatOpenAI
llm = ChatOpenAI(base_url="https://app.plynf.com/v1", api_key="plynf_sk_live_…", model="smart") LlamaIndex
Frameworkapi_base on the OpenAI LLM — RAG + agents route via Plynf.
from llama_index.llms.openai import OpenAI
llm = OpenAI(api_base="https://app.plynf.com/v1", api_key="plynf_sk_live_…", model="smart") Pydantic AI
FrameworkPoint the OpenAI model's base_url at Plynf.
from pydantic_ai.models.openai import OpenAIModel
model = OpenAIModel("smart", base_url="https://app.plynf.com/v1", api_key="plynf_sk_live_…") CrewAI
FrameworkSet two env vars — every crew + agent routes through Plynf.
export OPENAI_API_BASE="https://app.plynf.com/v1"
export OPENAI_API_KEY="plynf_sk_live_…" AutoGen
Frameworkbase_url in the model config — agents + group chats flow through.
config_list = [{
"model": "smart",
"base_url": "https://app.plynf.com/v1",
"api_key": "plynf_sk_live_…",
}] Haystack
Frameworkapi_base_url on the OpenAI generator.
from haystack.components.generators import OpenAIGenerator
gen = OpenAIGenerator(api_base_url="https://app.plynf.com/v1", model="smart") LiteLLM
GatewayFront your LiteLLM with Plynf — keep your routing, gain shaping.
model_list:
- model_name: smart
litellm_params:
model: openai/smart
api_base: https://app.plynf.com/v1
api_key: plynf_sk_live_… Cursor
Agent IDESettings → Models → Override OpenAI Base URL. Paste these two.
Override OpenAI Base URL: https://app.plynf.com/v1
OpenAI API Key: plynf_sk_live_… Claude Code
Agent IDEExport two vars — Claude Code routes through Plynf's Anthropic door.
export ANTHROPIC_BASE_URL="https://app.plynf.com"
export ANTHROPIC_API_KEY="plynf_sk_live_…" Cline
Agent IDEPick "OpenAI Compatible", paste the base URL + key.
Provider: OpenAI Compatible
Base URL: https://app.plynf.com/v1
API Key: plynf_sk_live_… Aider
Agent IDEOne flag (or env) and aider pairs through Plynf.
OPENAI_API_BASE=https://app.plynf.com/v1 OPENAI_API_KEY=plynf_sk_live_… \
aider --model openai/smart Windsurf
Agent IDEAdd an OpenAI-compatible provider in settings, paste these.
Base URL: https://app.plynf.com/v1
API Key: plynf_sk_live_… Zed
Agent IDESet the OpenAI api_url in Zed's assistant settings.
"language_models": {
"openai": { "api_url": "https://app.plynf.com/v1" }
} Open WebUI
Agent IDEAdmin → Settings → Connections → add an OpenAI connection.
API Base URL: https://app.plynf.com/v1
API Key: plynf_sk_live_… LibreChat
Agent IDEAdd a custom endpoint block to librechat.yaml.
endpoints:
custom:
- name: "Plynf"
baseURL: "https://app.plynf.com/v1"
apiKey: "plynf_sk_live_…"
models: { default: ["smart", "fast"] } n8n
AutomationCommunity node — connect with a key, drop into any workflow. No code.
Authorize once with your Plynf key — then use Plynf in any step. No code.
Zapier
AutomationPlynf app for Zaps — authorize once, use in any step.
Authorize once with your Plynf key — then use Plynf in any step. No code.
Make
AutomationPlynf module — drop it into any scenario.
Authorize once with your Plynf key — then use Plynf in any step. No code.
Copilot Studio
AutomationMicrosoft Copilot Studio custom connector for Plynf.
Authorize once with your Plynf key — then use Plynf in any step. No code.
Pipedream
AutomationIn any code step, use the OpenAI SDK pointed at Plynf.
import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://app.plynf.com/v1", apiKey: "plynf_sk_live_…" }); Ollama clients
LocalPoint any native-Ollama tool at Plynf's /api door — no rewrite.
export OLLAMA_HOST="https://app.plynf.com" Every other path
Self-host, managed deploys, team-chat bots and SDKs — ranked by friction.
1 · Run Plynf locally
Single command. Spins up the proxy + dashboard on your laptop in ~60 seconds. Best for evaluation, agent demos, and offline development.
- Requires Docker Desktop (or any Docker engine).
- Stop it again: `cd ~/.plynf && docker compose down`.
- Customise port: `… | sh -s -- --port 7431`.
2 · Deploy Plynf to a managed host
One click in your cloud provider of choice. We've shipped the right config files so you don't fiddle with Dockerfiles.
- Each button reads a config file in the repo and asks you only for required env vars (mostly empty for demo mode).
- Once deployed, point your bots/SDKs at the returned URL.
3 · Connect a no-code platform
Install Plynf inside platforms your team already uses. Each path takes ≤ 3 minutes.
Install community node
Settings → Community Nodes → Install → `n8n-nodes-plynf`
Install Zapier app
App directory → Plynf (link goes live after approval)
Install custom app
Make developer hub → Import → upload JSON files
Import as custom connector
Power Platform Maker → Import OpenAPI
4 · Add a Plynf bot to your team chat
Each bot is one-click deployable to Render, Fly or Railway. Bots talk to the proxy via the URL you set up above.
Add to Slack via OAuth
Run the OAuth handler then visit /slack/install
Bot Framework deploy
Register at dev.botframework.com → upload manifest in Teams
Slash command bot
Create app → register commands → npm start
SFDX deploy
`sf project deploy start`
5 · Use the SDK from your own agent code
Drop-in clients for the two languages people ship production agents in.
Plynf for Platforms
Build a platform agents run on? Let's make Plynf one click for your users.
We work directly with model providers, agent platforms and automation tools to ship native, one-click Plynf — co-built connectors, an "Add to Plynf" button in your product, pre-registered OAuth, and joint cost-savings reporting. Your users cut tokens 70%+ without leaving your surface.
We pair with your team to ship a native Plynf connector your users install in one click.
No dev-portal friction for end users — auth is brokered, scoped and revocable.
Show your users their token + cost savings in your product, powered by Plynf.
Already works with · OpenAI · Anthropic · Gemini · Cohere · Bedrock · Ollama · n8n · Zapier · Make · Slack · Salesforce
Stuck? Open an issue at github.com/nico-schindlbeck-jpg/plynf — we'll get you running in under 24 hours.