# 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.

Last verified 2026-10-19 · https://amux.ai/alternatives/litellm

## At a glance

| Option | Deployment | Pricing | Best for |
|---|---|---|---|
| Amux AI | Hosted | No markup or top-up fee; discounted low-cost and official channels | Teams that no longer want to run a proxy and want unified billing at lower cost with enterprise-grade organization management. |
| OpenRouter | Hosted | No inference markup; 5.5% fee on card top-ups | Developers who prioritize model count and need broad long-tail and open-weight coverage. |
| Requesty | Hosted | 5% service fee | Teams that need EU data residency, visual routing policies and an MCP gateway. |
| Vercel AI Gateway | Hosted | No markup; payment fees passed on | Teams already building on Vercel and the AI SDK. |
| Cloudflare AI Gateway | Hosted (edge network) | Core features free; 5% fee on Unified Billing credits | Teams on Cloudflare Workers that want caching and observability in front of existing provider accounts. |
| Portkey | Hosted or self-hosted | Priced by log volume, from $49/month | Enterprise teams that prioritize logging, guardrails and compliance. |

## What to look for

- **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.
- **Unified billing**: Whether one account and one balance cover every model, or you settle billing with each upstream provider separately.
- **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.
- **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.
- **Teams and governance**: Organizations, workspaces, role permissions and spend limits are the foundation for using LLMs across a growing team.
- **Data privacy**: Whether prompts and responses are stored, whether they are used for training, and whether zero-retention options exist.

## Top LiteLLM alternatives

### 1. Amux AI

A hosted LLM API gateway with 70+ 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.

### 2. OpenRouter

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. Requesty

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. Vercel AI Gateway

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. Cloudflare AI Gateway

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. Portkey

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

### Which LiteLLM alternative is right for me?

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.

### How do I keep data safe when moving from a self-hosted proxy to a hosted gateway?

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.

### Which costs less: a self-hosted proxy or a hosted gateway?

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.

### Is it hard to switch from LiteLLM to a hosted gateway?

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.

_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._