Amux AI

LiteLLM alternatives

The best LiteLLM alternatives

The leading LiteLLM alternatives are Amux AI, OpenRouter, Requesty, Vercel AI Gateway, Cloudflare AI Gateway and Portkey. Teams usually switch when they no longer want to maintain a self-hosted proxy: deployment, scaling, the database and cache are all on them, and every upstream needs its own account and top-ups. Below, six leading options are compared on hosting model, pricing, protocol compatibility and governance.

Get startedBrowse models and pricing

Last verified 2026-10-19

At a glance

OptionDeploymentPricingBest for
Amux AIOur productHostedNo markup or top-up fee; discounted low-cost and official channelsTeams that no longer want to run a proxy and want unified billing at lower cost with enterprise-grade organization management.
OpenRouterHostedNo inference markup; 5.5% fee on card top-upsDevelopers who prioritize model count and need broad long-tail and open-weight coverage.
RequestyHosted5% service feeTeams that need EU data residency, visual routing policies and an MCP gateway.
Vercel AI GatewayHostedNo markup; payment fees passed onTeams already building on Vercel and the AI SDK.
Cloudflare AI GatewayHosted (edge network)Core features free; 5% fee on Unified Billing creditsTeams on Cloudflare Workers that want caching and observability in front of existing provider accounts.
PortkeyHosted or self-hostedPriced by log volume, from $49/monthEnterprise teams that prioritize logging, guardrails and compliance.

What to look for

  1. 01

    Hosted or self-hosted

    Hosted services remove deployment and operations work; self-hosted options give full control of the data path but leave upstream accounts, keys and scaling to you.

  2. 02

    Unified billing

    Whether one account and one balance cover every model, or you settle billing with each upstream provider separately.

  3. 03

    Pricing and fees

    Whether inference is marked up, whether top-ups carry a fee and whether credit expires. At high volume, a few percentage points translate directly into cost.

  4. 04

    Protocol compatibility

    Whether the protocols your SDKs and tools use are supported natively. Native support means migrating by changing the base URL, with no code rewrite.

  5. 05

    Teams and governance

    Organizations, workspaces, role permissions and spend limits are the foundation for using LLMs across a growing team.

  6. 06

    Data privacy

    Whether prompts and responses are stored, whether they are used for training, and whether zero-retention options exist.

Top LiteLLM alternatives

  1. 01

    Amux AI

    Our productHosted

    A hosted LLM API gateway with 80+ text, image and video models, unified billing in one account, no inference markup, no top-up fees and low-cost channels far below official pricing and discounted official channels. Natively supports OpenAI, Anthropic and Gemini protocols plus vendor image and video protocols.

    Pricing
    No markup or top-up fee; discounted low-cost and official channels
    Best for
    Teams that no longer want to run a proxy and want unified billing at lower cost with enterprise-grade organization management.
    Considerations
    Embeddings, speech and SAML SSO are on the roadmap; bringing your own upstream keys is not supported yet.
    Get startedSee the full Amux AI vs LiteLLM comparison
  2. 02

    OpenRouter

    Hosted

    A hosted model marketplace advertising 500+ models from 80+ providers, with BYOK and enterprise SSO.

    Pricing
    No inference markup; 5.5% fee on card top-ups
    Best for
    Developers who prioritize model count and need broad long-tail and open-weight coverage.
    Considerations
    Top-ups carry a fee; workspace budgets and role assignment are Enterprise-only.
  3. 03

    Requesty

    Hosted

    A hosted, OpenAI-compatible router advertising 600+ models, with fallbacks, load balancing and latency routing.

    Pricing
    5% service fee
    Best for
    Teams that need EU data residency, visual routing policies and an MCP gateway.
    Considerations
    Adds a 5% service fee; group budgets, full RBAC and SSO are Enterprise-only.
  4. 04

    Vercel AI Gateway

    Hosted

    Vercel's hosted gateway with no markup and no platform fee on tokens, including BYOK.

    Pricing
    No markup; payment fees passed on
    Best for
    Teams already building on Vercel and the AI SDK.
    Considerations
    Requires a Vercel team account; purchased credits expire after one year.
  5. 05

    Cloudflare AI Gateway

    Hosted (edge network)

    A gateway on Cloudflare's edge network with free analytics, caching and rate limiting; credits are purchased through Unified Billing.

    Pricing
    Core features free; 5% fee on Unified Billing credits
    Best for
    Teams on Cloudflare Workers that want caching and observability in front of existing provider accounts.
    Considerations
    Unified Billing charges a 5% fee on credit purchases; log allowances depend on the Workers plan.
  6. 06

    Portkey

    Hosted or self-hosted

    A gateway centered on observability and governance, available as open-source self-hosted software or a hosted service.

    Pricing
    Priced by log volume, from $49/month
    Best for
    Enterprise teams that prioritize logging, guardrails and compliance.
    Considerations
    Priced by log volume — the Production plan is $49/month for 100,000 logs; you bring your own upstream accounts.

Frequently asked questions

It depends on your priority. To drop operations work, get unified billing and lower costs, choose Amux AI. For the largest model catalog, choose OpenRouter. For EU data residency, choose Requesty. If you are deep in the Vercel or Cloudflare ecosystem, use that platform’s gateway. If you still want to self-host but care most about observability and governance, choose Portkey.

Check three things: whether prompts and responses are stored, whether they are used for training, and what is kept when a call fails. Amux AI does not store bodies of successful text calls, never trains on customer content, and keeps failure samples for at most 15 days, which you can turn off.

A self-hosted proxy has no platform fee, but you pay each upstream at its official price and carry servers, databases and operations staff. Among hosted gateways, Amux AI has no inference markup and no top-up fee, with low-cost channels far below official pricing and discounted official channels, so at most volumes the total cost is lower.

Most hosted gateways are compatible with the OpenAI SDK, so switching usually means changing the base URL and key, and replacing the custom model names from your proxy with the target platform’s model IDs.

Get started

Minutes from now,
your first request is live.

  1. 1Create an account
  2. 2Create an API key
  3. 3Swap the base_url and call
Base URL
Get startedView pricing

More comparisons

LiteLLM and other product names mentioned on this page are trademarks of their respective owners. Information about third-party products is compiled from their official public sources for evaluation purposes only and may change; refer to each vendor's website for current details. To report an inaccuracy, contact support@amux.ai.