YourAITools

Best LLM routers and AI gateways: one API for many models

Updated Nov 6, 2026. Listed A to Z, not ranked. Prices change often: confirm with each provider.

An LLM router (or AI gateway) puts many model providers behind one OpenAI-compatible endpoint, with fallbacks, routing and spend tracking. The main pricing differences are how the router earns: a fee on credits (OpenRouter), a markup on model cost (Requesty), a platform subscription (Portkey) or nothing, if you self-host open source (LiteLLM).

Below: published fees from each official page, then the routers in our directory that agents can pay per call in USDC.

Router and gateway fees

ProviderPublished pricingNotes
LiteLLMOpen-source gateway and Python SDK (self-host)One API to 100+ LLM providers, MCP tools and A2A agents; keys, budgets and per-request cost tracking
OpenRouterFree (25+ free models, 50 requests/day); Standard 5.5% fee on credit purchases; Business 8%; Enterprise customBYOK: $25,000/month of list-price inference with no fee, then 5%
PortkeyDeveloper free (10k logs/month); Production $49/month (100k logs, +$9 per extra 100k); Enterprise customOpen-source gateway also available to self-host
RequestyFree (free models, 200 requests/day); pay as you go 5% markup on model cost; Enterprise customNo per-seat pricing or minimum spend; 600+ models

Quoted from each provider’s official pricing page on Nov 6, 2026.

Routers and gateways in our directory (9)

ToolWhat it doesHow you payPublished priceLinks
BlockRunPay-per-result access to AI models and tools for agents, with each payment tracked and scored so spend goes to what works.USDC per call (x402)$0.002–$0.0085 per call (listing) (checked 2026-10-22)Listing
GondolaOpenAI- and Anthropic-compatible endpoint for frontier models paid in USDC, without storing prompts.USDC per call (x402)$0.01–$0.011134 per call (listing) (checked 2026-10-22)Listing
JarvisClawAI API gateway with one OpenAI-compatible endpoint for GPT, Claude, Gemini and other models, with smart routing and pay-per-call via x402 in USDC on Base or Solana.USDC per call (x402)$0.003–$0.00414 per call (listing) (checked 2026-10-22)Listing
LosbetoOpenAI-compatible LLM gateway paid per call in USDC, plus market data such as forex, equities, commodities and Brazilian macro.USDC per call (x402)$0.005–$0.99 per call (listing) (checked 2026-10-22)Listing
MapleAIOpenAI-compatible chat (GPT and Claude models), image generation and editing, and audio behind one endpoint, paid per request in USDC over x402. No subscription or API key.USDC per call (x402)From $0.001 (varies by model and tokens) per call (listing) (checked 2026-10-27)Listing
OpenRouterUnified API for hundreds of LLMs; credits can be bought with cryptocurrency (crypto payments carry a fee and are non-refundable per its FAQ).USDC per call (x402)See provider (checked 2026-10-21)Listing · Pricing
Synap by LinkrraOpenAI-compatible inference API for popular open models. Pay per call in USDC on Base via x402, no account needed, or use a prepaid API key.USDC per call (x402)From $0.001 (varies by model) per call (listing) (checked 2026-10-25)Listing
x402engineOver 100 APIs behind x402 payments, including LLMs, video and image generation, code execution, travel search and TTS.USDC per call (x402)See provider (checked 2026-10-22)Listing
ZerouterAI API gateway with one key for Claude, GPT, Gemini, Grok and Kimi models, plus Qwen, DeepSeek and GLM via Alibaba Cloud, over OpenAI- and Anthropic-compatible endpoints that work with Claude Code, Codex CLI and Cursor. Usage is billed per million tokens from a prepaid balance that can be topped up with USDC or USDT from a crypto wallet, or by card. Per its terms, it is a marketplace that routes calls to official AI accounts whose owners rent out capacity, and all sales are final.USDC per call (x402)From $0.007 per 1M input tokens (varies by model) per call (listing) (checked 2026-11-02)Listing

How to compare router fees

  • Percentage fees scale with spend: OpenRouter’s 5.5% is charged when you buy credits; Requesty’s 5% is added to model cost ($10 of tokens costs $10.50).
  • Subscriptions (Portkey) bill for logs and features, not tokens; you still pay providers directly.
  • Self-hosted (LiteLLM, Portkey open source) have no router fee, but you run and secure the gateway.
  • Per call in USDC: directory routers such as Zerouter, MapleAI and Linkrra accept x402 or crypto top-ups, so an agent can pay without a card.

See OpenRouter pricing, OpenRouter alternatives and Venice AI vs OpenRouter.

FAQ

What is an LLM router?
A service or self-hosted gateway that exposes many model providers through one API, adding routing, fallbacks, caching and spend tracking.
How much does OpenRouter charge?
On openrouter.ai/pricing (Nov 6, 2026): a 5.5% fee on credit purchases on Standard and 8% on Business; the Free tier allows 50 requests a day on free models.
Is there a free LLM router?
LiteLLM and Portkey’s gateway are open source and free to self-host; Portkey’s hosted Developer plan and Requesty’s Free tier are free with limits.

Sources (read Nov 6, 2026): openrouter.ai/pricing · www.requesty.ai/pricing · portkey.ai/pricing · docs.litellm.ai/

Picking tools for an agent? Ask other builders + take the free x402 course

Join the community

Keep exploring

Guides