Agentic WeeklyGet The Weekly Verdict, freeSubscribe free

OpenRouter vs LiteLLM

As of Nov 3, 2026 (CT), OpenRouter is a hosted API that bills each model at the provider's list price, plus a 5.5% fee on credit purchases (8% on Business). Free models are limited to 50 requests a day. LiteLLM is an open-source gateway and Python SDK, MIT outside its enterprise folder, that you host yourself. It calls 100+ LLMs in the OpenAI format with your own keys.

Our pick: Pick OpenRouter for one hosted key and one bill across 600+ models with no servers to run; pick LiteLLM to keep traffic in your own infrastructure, use your own provider keys and pay no gateway fee beyond your hosting. Updated (America/Chicago).

By Agentic Weekly, operated independently from Texas, US · Last updated (America/Chicago) · How we work

Last verified Nov 3, 2026 (America/Chicago) on OpenRouter pricing, LiteLLM docs, the LiteLLM licence and the GitHub and PyPI APIs. Evidence: Mixed. Our LiteLLM smoke test on Oct 25, 2026 is Tested within its scope (Python SDK, local 3B model); OpenRouter is Reported, and we have not run it.

OpenRouter prices: openrouter.ai/pricing ·

OpenRouter

OpenRouter

Reported

Unified API to many models with fallbacks; a Free tier with 25+ free models and 4 free providers.

LiteLLM

BerriAI · v1.106.4

Tested (smoke test)

Useful as a single interface in front of local and hosted models; check the license files and pin the version because it releases daily.

Side by side, from what we recorded

Each row compares one recorded fact for both tools. Rows are not weighted and nothing is added up into a score; the note under each row says what the marker means.

  • Evidence

    Reported for OpenRouter
    Tested (smoke test) (LiteLLM is ahead on this row) for LiteLLM

    Tested = we ran it (smoke test); Reported = we read official sources

  • Pricing we could verify

    Verified from official page (OpenRouter is ahead on this row) for OpenRouter
    Partly verified for LiteLLM

    About our evidence only, not about the price

  • Steps in our setup guide

    2 for OpenRouter
    2 for LiteLLM

    Counts our guide's steps, not difficulty

  • Sources we cite

    3 for OpenRouter
    2 for LiteLLM

    How much we had to read; more is not better

Recorded differences, not a quality score; any head-to-head results we ran are in their own section below. How we label evidence.

Hosted router vs self-hosted gateway

OpenRouter is a service: you buy credits from OpenRouter (or bring your own provider keys), call one API, and OpenRouter forwards each request to a model provider. LiteLLM is software: a Python SDK you import, or a proxy server ("LLM Gateway") you deploy, with virtual keys, cost tracking and an admin UI. With LiteLLM, each provider bills you directly and you pay for the machine the proxy runs on.

OpenRouter vs LiteLLM (sources read Nov 3, 2026, CT)
OptionWhat it isWho hosts itLicensePrice of the software/serviceFee on model usageBYOKFree tier limitsProviders/models (vendor claim)Our evidence label
OpenRouterHosted multi-provider APIOpenRouterProprietary serviceFree, Standard and Business are pay-as-you-go; Enterprise by contractNone per request; 5.5% on credit purchases (Standard), 8% (Business), discounts on Enterprise$25,000 of list-price inference a month with no fee, then 5% (Standard and Business)Free plan: 25+ free models, 4 free providers, 50 requests/day600+ models on Standard and BusinessReported
LiteLLMOpen-source Python SDK and self-hosted gateway (proxy)YouMIT, except the enterprise/ directory (own licence); GitHub shows NOASSERTION$0 for the open-source software; Enterprise price not publishedNone from LiteLLM; you pay providers and your hostingAlways: you use your own provider keysNo LiteLLM-imposed limits; provider limits apply"100+ LLMs" in the OpenAI formatTested Oct 25 (SDK only); rest Reported

LiteLLM's latest release is v1.106.6, published Oct 28, 2026 (CT); the repo had 62,632 GitHub stars on Nov 3. Sources: OpenRouter pricing, LiteLLM docs, releases.

Cost side by side

  • OpenRouter: "Inference is billed at the provider's list price on every plan." Its fee is charged when you buy credits, "never on individual requests". With BYOK, the provider bills you directly and Standard and Business include $25,000 of list-price inference a month with no OpenRouter fee, then a 5% fee.
  • OpenRouter Enterprise: pricing set by contract, based on volume, prepaid credits and annual commitments, with invoicing.
  • LiteLLM: the open-source SDK and proxy cost nothing; you pay each provider at its own rates plus whatever your server costs. LiteLLM sells an Enterprise tier, but no price was published on the pages we read, so we do not state one.
  • For model prices themselves, see LLM pricing comparison and cheapest LLM API.

Worked example: $100 of model usage

Our arithmetic, not a vendor quote. To spend $100 of list-price inference through OpenRouter Standard, you buy $100 of credits plus the 5.5% fee: $105.50. On Business at 8%, $108. Through LiteLLM, the same usage is $100 billed by the providers, plus the cost of hosting the proxy, which depends on your setup. Under OpenRouter BYOK, the first $25,000 of list-price inference a month carries no OpenRouter fee.

Limits and reliability

OpenRouter's Free plan is capped at 50 requests a day; on Standard and Business, rate limits are "passthrough from the provider(s)". Per-minute caps on free models were not stated on the pricing page we read on Nov 3, so we do not give one.

LiteLLM's docs list a Router "with retry, fallback, and load balancing" across deployments, which you configure yourself; provider rate limits still apply.

Data handling

With LiteLLM, requests pass through infrastructure you run before reaching the provider. With OpenRouter, they pass through OpenRouter. OpenRouter's data-retention options were not on the pricing page we read, so check its own documentation before relying on any routing guarantee.

What we ran (Oct 25, 2026)

Our LiteLLM smoke test (Tested, scope only)
VersionTaskResultMedian timeWhat went wrong
LiteLLM 1.106.4, Python SDKCall one tool (add 17 + 25) through a local Qwen2.5-3B modelPassed 3 of 36.7 s (trials 5.3 to 7.2 s)A plain install failed to import: No module named 'mcp'; installing mcp 2.3.0 fixed it

We did not test the LiteLLM proxy, budgets, keys or routing, and we have not run OpenRouter. Full log: LiteLLM test.

Which to pick when

  • Pick OpenRouter if you want one key, one bill and 600+ models without running a server, and a 5.5% fee on credits is acceptable.
  • Pick LiteLLM if you already hold provider keys, want traffic inside your own infrastructure, or want per-team virtual keys and spend tracking you control.
  • Use both if you want LiteLLM's gateway in front of OpenRouter: LiteLLM's provider list includes OpenRouter (openrouter) (LiteLLM OpenRouter docs).
  • Running models locally instead? See Ollama. More options: OpenRouter alternatives, best LLM platforms.

FAQ

Does OpenRouter mark up model prices?

OpenRouter says no: inference is billed at the provider's list price on every plan, and its fee (5.5% on Standard, 8% on Business) is charged when you buy credits.

Is LiteLLM free?

The open-source SDK and proxy are free under MIT, except the enterprise/ directory. You pay your model providers and your hosting; LiteLLM's Enterprise price is not published.

Can LiteLLM call OpenRouter?

Yes, LiteLLM lists OpenRouter as a provider (openrouter) in its docs, read Nov 3, 2026.

What is the latest LiteLLM version?

v1.106.6, published Oct 28, 2026 (CT), per GitHub releases read Nov 3, 2026. We tested 1.106.4 on Oct 25.

Sources

Page history

  • : Rewritten in place: answer first, a dated comparison table and sources, with every price re-read on the vendor pages on Nov 3, 2026 (America/Chicago). The page was noindexed before this rewrite.

Markdown version for agents: /compare/openrouter-vs-litellm.md