If you want LiteLLM without running it, there is no LiteLLM-run hosted service listed on its site or docs: the managed alternatives are INFRO, Portkey, Cloudflare AI Gateway and Vercel AI Gateway, and OpenRouter if you would rather not hold provider accounts. If you want to keep self-hosting, Bifrost and Portkey's open-source gateway are the closest swaps; either way, the March 2026 incident is a reason to tighten your supply chain, not only to change products.
Not one LiteLLM runs for you, as far as its own pages show. Its site describes the gateway as "Open source. Self-hosted.", deployed into your own cloud, and its Enterprise tier is a license key on the same image you run. Third-party templates (Railway, Render and similar) start a LiteLLM container on your account; you still own its upgrades and its database.
So "hosted LiteLLM" means one of two things in practice: LiteLLM on infrastructure you rent, which is still self-hosting, or a managed gateway that does the same job, which is what the rest of this page compares.
On 24 March 2026, two LiteLLM releases on PyPI (1.82.7 and 1.82.8) carried credential-stealing code for about 40 minutes. LiteLLM traced the compromise to a dependency in its CI pipeline, removed the packages, and released a clean 1.83.0; it says customers running the official LiteLLM Proxy Docker image were not affected.
The lesson applies to every product here, hosted or not, because a gateway holds a key for every provider account you have:
Install by hash, not by version range, and prefer images with pinned dependencies.
Pin CI actions to commits, not tags.
Give the gateway inference-only keys, one per provider account, never an admin key.
Rehearse rotating every key it holds, and time it.
Ask a hosted vendor how it builds, signs and deploys what it runs.
Each option is described from its own pricing page, docs or repository, checked on the date at the top. INFRO is listed first because this is INFRO's site; its limits are listed as plainly as everyone else's.
Teams that want LiteLLM's job (one API, budgets, failover on their own keys) without running the service.
How it charges
A flat plan of $35–$599 a month, chosen by request volume (100K–10M requests a month), never by spend. 0% markup on model spend: your providers bill you directly.
Strengths
Every request runs on a provider account you own; INFRO holds none, so nothing falls back onto someone else's contract.
Budgets refuse a request before the provider is called, per organization, project, member and key, and (from the Team plan) per end customer.
Each response names the provider, connection and model that served it, with an estimated provider cost.
Limits
No shared capacity: when every connection you own fails, the request fails with an error that says why.
Hosted only on its plans; a self-hosted deployment needs an enterprise agreement.
Enterprises buying an AI gateway as part of a security platform, or teams self-hosting Portkey's open-source gateway.
How it charges
Production is $49 a month with 100,000 recorded logs, plus $9 per additional 100,000. The open-source gateway is MIT licensed. Enterprise is custom.
Strengths
Mature routing, fallbacks, guardrails and observability.
An MIT-licensed gateway you can host yourself.
Limits
Palo Alto Networks completed its acquisition on 29 May 2026; the product is now Prisma AIRS AI Gateway, so check its roadmap fits a team not buying the wider platform.
Logs, observability and prompt management are in the hosted app, not the open-source gateway.
Stay when prompts and keys must not leave your network, you already run it well, and you have the people to keep it patched. Pin a verified version by hash, prefer the pinned Docker image, isolate the proxy with only the secrets it needs, and keep up with its security releases. For comparison, INFRO vs LiteLLM sets out what each costs you to run.
Its own site describes it as open source and self-hosted, deployed into your cloud, and no LiteLLM-run hosted service was listed when we checked. Managed gateways such as INFRO, Portkey, Cloudflare and Vercel do the same job without you running it.
Is LiteLLM safe to use after the supply-chain attack?
LiteLLM removed the two affected PyPI releases, moved to a new pipeline and says its official Proxy Docker image was not affected. Pin verified versions by hash and isolate the proxy, as you should for any gateway that holds provider keys.
What is the closest alternative to the LiteLLM proxy?
To keep self-hosting: Bifrost or Portkey's open-source gateway. To stop running it: a hosted BYOK gateway such as INFRO, which keeps every request on your own provider accounts.
Sources
Every statement about another company on this page comes from that company's own page, opened on the date shown. Prices and defaults change; check the source before you decide.
On 24 March 2026, litellm 1.82.7 and 1.82.8 on PyPI carried credential-stealing code for about 40 minutes; LiteLLM says the official Proxy Docker image was not affected and 1.83.0 is clean.
Vercel AI Gateway charges no markup and no platform fee on tokens; you pay the provider's list price.
Bringing your own provider key adds no Vercel fee.
When a request on your own credentials fails, Vercel retries it on its own system credentials and charges that usage to your Vercel AI Gateway balance.
Spend on your own credentials is not counted in Vercel AI Gateway budgets, so a budget cannot cap BYOK spend.
Past the allowance, BYOK usage is charged 5% of what the same model and provider would cost on OpenRouter.
By default, a BYOK request that fails or is rate limited on your key falls back to OpenRouter's shared endpoints, paid from your OpenRouter balance; a per-key "Shared capacity fallback" setting turns that off.
The team building INFRO, a BYOK-first control plane for production AI. Articles are written from the engineering work on the gateway and cite their external sources.