COMING SOON
20+
providers
60+
models
$0
monthly cost

Free AI is everywhere.
Nobody's using it all.

Gemini, Groq, Mistral, Cerebras, SambaNova, OpenRouter — dozens of providers offer free tiers. When those run out, DeepSeek and Alibaba charge fractions of a cent. Freeloader exhausts every free tier first, then routes to the cheapest option.

FREELOADER
Terminal — freeloader
✓ OpenAI-compatible API
✓ Works with any SDK
✓ MIT licensed
✓ Self-hosted & private

The problem isn't AI.
It's paying for it.

You call GPT-4o. You pay $2.50/M tokens. Meanwhile, Gemini Flash does the same thing for free. So does Groq. So does Cerebras. You just need something that knows this — and knows what to do when free runs out.

YOUR APP
OPENAI_BASE_URL=localhost:4011
FREELOADER PROXY
Checks quotas → routes to cheapest available → falls back automatically
1. FREE TIER FIRST
2. LOWEST COST NEXT
Gemini
FREE
Groq
FREE
Mistral
FREE
OpenRouter
FREE
Cerebras
FREE
SambaNova
FREE
Ollama
LOCAL
DeepSeek
BUDGET
Alibaba
BUDGET
Together
BUDGET
Cohere
BUDGET
OpenAI
Anthropic
+more

Free tiers first. Then cheapest paid. Premium only as last resort.

Pricing.

We agonized over this for weeks.

Hobby
$0
/mo forever
  • ✓ 20+ providers, 60+ models
  • ✓ Unlimited requests
  • ✓ Smart fallbacks
  • ✓ Trust tiers
  • ✓ Dashboard & reports
yes, all of it. for free.
MOST POPULAR
Pro
$0
/mo forever
  • ✓ Pretty much the same thing
we just wanted three columns
Enterprise
$0
/mo forever
  • ✓ Maybe you're not understanding the concept of free
MIT licensed. read it and weep.
Instruction Manual
$ npx xswarm-freeloader
that's pretty much it.

"But what about my sensitive data?"

Fair question. Freeloader has trust tiers. Tag your requests and control exactly where your data goes.

OPEN

Anything goes

Blog posts, summaries, translations — non-sensitive work routed to any provider for maximum free tier coverage.

Gemini Groq Cerebras OpenRouter Mistral +more
All providers
STANDARD

Trusted providers

Customer data, internal tools — routed only to providers with SOC 2 / GDPR compliance. Still mostly free.

Gemini Mistral OpenAI Anthropic
Trusted only
PRIVATE

Never leaves your machine

PII, medical data, secrets — routed exclusively to local models. Ollama, LM Studio, llamafile. Zero network calls.

Ollama LM Studio llamafile
Local only

It gets better.

LOCAL MODELS

Got a GPU? Add it to the mix.

Running Ollama? LM Studio? llamafile? Add them as providers. Freeloader routes to them like any other — except they're free, fast, and nothing ever leaves your machine.

$ ollama pull llama3.2
// freeloader detects it automatically
REPORTS

Show your boss the savings.

Daily email digests and PDF savings reports. Show exactly how much you would have spent vs. what you actually spent. Your finance team will send you a fruit basket.

Daily digests PDF reports Cost comparison
PRIVACY

All local. All yours.

SQLite database on your machine. No cloud account. No telemetry. No phone-home. Your API keys, your data, your ~/.xswarm/ directory. Runs behind your firewall. Uninstall with one command.

ECOSYSTEM

Part of xSwarm.

Freeloader is the free tool in the xSwarm family. xSwarm manages teams of Claude Code projects with remote monitoring, planning, and reporting. Buzz uses agent swarms for aggressive guerrilla outreach for your SaaS.

Three steps. Zero dollars.

Works with every OpenAI SDK. LangChain. Vercel AI. Your own fetch calls. If it speaks OpenAI, it speaks Freeloader.

1

Install & run

$ |

Sets up pm2, SQLite DB, syncs 127 models, starts router + dashboard.

2

Configure free tiers

Dashboard → http://localhost:4010

Add your free-tier API keys for Gemini, Groq, Mistral, and more.

3

Point your app

OPENAI_BASE_URL=http://localhost:4011

That's it. Your existing code stays the same. Your bill doesn't.

app.js — your existing code, unchanged
import OpenAI from 'openai';
 
const client = new OpenAI({
baseURL: 'http://localhost:4011',
apiKey: 'free', // any string works
});
 
// asks for gpt-4o, gets Gemini Flash (free)
const res = await client.chat.completions.create({
model: 'gpt-4o', // aliased to free tier
messages: [{ role: "user", content: "..." }]
});
100%

Free Tier Rotation

Rotates requests across Gemini, Groq, Cerebras, Mistral, SambaNova, OpenRouter, and local models — auto-throttling each to stay within free limits.

$$$

Smart Fallbacks

Circuit breakers, health checks, exponential backoff. If a provider hiccups, your request lands somewhere else in milliseconds.

$0

Dashboard

Seven views: live feed, provider health, usage analytics, budget tracking, routing config, opportunities, settings. All local.

CAP

Budget Hard Caps

Set $0/month. Or $5. Or $50. Freeloader enforces it per-app, per-project. Your credit card rests easy.

∞

Local Models

Add any local LLMs — Ollama, LM Studio, llamafile — behind a single router endpoint. Private data never leaves your machine.

600+

Key Sanitization

Every request is scrubbed for 600+ API key patterns using xswarm-ai-sanitize to help prevent accidental secret leakage.

Start saving.

New free tiers appear constantly. We'll keep you updated on providers, models, and releases.

New providers, free tier changes, and releases. No spam, ever.

Or just run npx xswarm-freeloader right now. It's free. Obviously.