We may earn a commission if you buy through links on this page. It never changes a score or a ranking. How we make money
LiteLLM is an open-source, self-hostable gateway for routing requests across model providers. Developers choosing a replacement can trade control of the gateway for a hosted API, or keep self-hosting with another open-source option. Reddit threads reported a compromise involving LiteLLM 1.107.0 and its Docker image, including credential-stealing behavior. For any gateway that holds provider keys, pin releases, review the release pipeline and confirm how you will rotate credentials before moving production traffic.
01Why people leave LiteLLM
Release and supply-chain risk
Reddit threads reported a compromise involving LiteLLM 1.107.0 and its Docker image. Pin exact releases and review credentials exposed to affected installations.
Credential blast radius
A gateway centralizes provider access. A compromised host or release can put multiple API keys at risk, so teams may prefer a different operational boundary.
Different hosting tradeoffs
Some teams want a hosted routing API; others need self-hosting. Compare deployment control, pricing and API compatibility against your production requirements.
02The alternatives, ranked by fit
- Best for AI SDK teams
Vercel AI GatewayVercel AI GatewayVercel AI Gateway is a hosted model-routing API for teams already using the AI SDK. Its free plan is listed, but production costs and limits need checking against your traffic. You avoid operating a gateway yourself; in exchange, you depend on a hosted service. Start migration with one workload and verify provider routing and failure behavior before moving more traffic.
Trade-off: Hosted service; confirm production pricing, limits and routing behavior.
7.8/10from
Freefree plan - Best for model choice
OpenRouterOpenRouterOpenRouter offers one API for models across providers, with provider failover among its stated use cases. That can reduce the provider-specific integration work in a multi-model setup. A free plan is listed, but compare the cost of your actual model mix and check routing before production. It is a hosted option rather than a self-hosted replacement.
Trade-off: Hosted routing means less control over the gateway infrastructure.
7.6/10from
Freefree plan - Best self-hosted gateway
PortkeyPortkeyPortkey is an open-source, self-hostable gateway for teams that want centralized model access and operations without relying on LiteLLM. A free plan is listed, with pricing from $49; confirm the applicable plan and deployment costs for your use. Migration means adopting Portkey’s gateway and checking its API and controls against your existing clients.
Trade-off: Verify production pricing and private-deployment details before committing.
7.6/10from
$49/mofree plan - Best for Cloudflare teams
Cloudflare AI GatewayCloudflare AI GatewayCloudflare AI Gateway is a hosted option for teams already using Cloudflare. Its stated features include multi-provider routing, caching and AI request logs, which can consolidate gateway operations without self-hosting another service. A free plan is listed, but confirm pricing and limits for production. Test your request paths and logging needs before redirecting existing traffic.
Trade-off: Hosted service; pricing and production limits need confirmation.
7.0/10from
Freefree plan - Best for self-hosted visibility
HeliconeHeliconeHelicone combines LLM routing and observability and is listed as open source and self-hostable. It may suit teams that want request-level cost visibility while retaining control of deployment. A free plan is listed, with pricing from $79. Before adopting it, verify current maintenance status and whether its routing and operational model fit your workloads.
Trade-off: Check maintenance status and the cost of your intended plan.
6.8/10from
$79/mofree plan - Best for managed fallback
RequestyRequestyRequesty is a hosted gateway for routing across model providers, with fallback as a stated use case. It may replace custom retry logic with a central service, but confirm how its routing behaves for your models and failure cases. A free plan is listed; verify service limits and costs using expected production volume before switching.
Trade-off: Hosted service; confirm costs, limits and fallback behavior.
7.5/10from
Freefree plan - Best for OpenAI-compatible access
Eden AIEden AIEden AI is a hosted OpenAI-compatible gateway with cost and access controls, and its listed starting price is $0. It may reduce client changes where your integration uses the compatible API, but validate the endpoints and behavior your application needs. The vendor does not publish a self-hosting path, so this is not a fit when deployment control is required.
Trade-off: No published self-hosting path; confirm production costs and limits.
7.2/10from
Freefree plan
03Side by side
| Tool | Score | From | Free plan | Best for |
|---|---|---|---|---|
 LiteLLM LiteLLM (current) | 7.1 | Free | Yes | Platform teams managing multiple model providers |
| Vercel AI Gateway Vercel AI Gateway | 7.8 | Free | Yes | Teams integrating multiple model providers |
| OpenRouter OpenRouter | 7.6 | Free | Yes | Teams evaluating models across providers |
| Portkey Portkey | 7.6 | $49/mo | Yes | Teams routing requests across LLM providers |
| Cloudflare AI Gateway Cloudflare AI Gateway | 7.0 | Free | Yes | Teams already using Cloudflare |
| Helicone Helicone | 6.8 | $79/mo | Yes | Teams routing across model providers |
| Requesty Requesty | 7.5 | Free | Yes | Teams routing across multiple model providers |
| Eden AI Eden AI | 7.2 | Free | Yes | Teams routing across AI providers |
Questions people ask
Which LiteLLM alternatives can I self-host?
Portkey and Helicone are listed as open source and self-hostable. That makes them the clearest options here if you need to operate the gateway in your own environment. Check each project’s current deployment documentation, release process and production requirements before moving provider credentials.
Which alternative has an OpenAI-compatible API?
Eden AI is described as a hosted OpenAI-compatible gateway. Compatibility does not guarantee that every endpoint or behavior your application uses will match, so validate your request and response paths before migrating. The other options should be checked against their API documentation for your specific client needs.
How should I migrate away from LiteLLM safely?
Start with one non-critical workload and compare API behavior, provider routing and failure handling before shifting production traffic. Pin exact versions of any self-hosted gateway, avoid unreviewed automatic updates, and rotate credentials if they may have been exposed. Compare costs using your actual model mix and request volume.