Comparison
There are many LLM gateways. FreeLLM occupies a position none of them target: truly $0 by design, built specifically for stitching free-tier providers into a reliable dev tier.
| Feature | FreeLLM | LiteLLM | OpenRouter | Portkey |
|---|---|---|---|---|
| Truly $0 (no markup) | ✅ | Self-host | ❌ 5% markup | ❌ $49+/mo |
| Multi-key rotation per provider | ✅ | ❌ | ❌ | ❌ |
| OpenAI-compatible | ✅ | ✅ | ✅ | ✅ |
| Automatic failover | ✅ | ✅ | ✅ | ✅ |
| Built-in real-time dashboard | ✅ | ❌ | ✅ | ✅ |
| Per-provider token tracking | ✅ | ✅ | ✅ | ✅ |
| Circuit breakers | ✅ | Partial | ✅ | ✅ |
| Self-hosted | ✅ | ✅ | ❌ | Both |
| TypeScript codebase | ✅ | ❌ Python | ? | ❌ |
| One-click cloud deploy | ✅ | ❌ | n/a | ❌ |
When to use which
- FreeLLM. You want production-quality reliability for $0. You’re prototyping, building a side project, or running internal tools.
- LiteLLM. You need 100+ provider support and don’t mind YAML config + Python.
- OpenRouter. You want zero ops effort and don’t mind paying ~5% markup on tokens.
- Portkey. You’re an enterprise team with budget and need SSO, audit logs, RBAC, and guardrails.
What FreeLLM is NOT trying to be
We deliberately don’t compete with LiteLLM on provider count, OpenRouter on hosted convenience, or Portkey on enterprise features. The “free tier” angle is the moat. Every feature in FreeLLM answers one question: Does this help someone get more value out of free LLM APIs?
A note on supply chain security
In late 2025, LiteLLM had a significant supply chain attack. Malicious versions on PyPI shipped credential stealers and persistent backdoors.
If you’re concerned about LLM gateway supply chain risk, FreeLLM has structural advantages:
- ~2k lines of code vs LiteLLM’s 100k+. Auditable in an afternoon.
- TypeScript instead of Python. Different dependency ecosystem.
- Pinned dependencies with
pnpm-lock.yamlcommitted - Multi-arch Docker images signed and published via GitHub Actions to GHCR
- Single-developer repository with traceable commits
Pick the gateway that matches your threat model.