Guides·Gateways

LLM gateways and routing

Every serious AI setup eventually puts one endpoint in front of many providers. These guides explain the mechanics, then compare the gateways you can actually run, including the one built into Continuum.

Download for Mac All guides
01

Start here

Gateways

What is OpenRouter? One key, 400+ models, and what it costs you

One API key in front of every model vendor. What it costs, what it hides, and the two cases where going direct is better.

Read guide →8 min Gateways

What is an LLM gateway? One endpoint in front of every model

One endpoint, many providers. What a gateway actually does, what it costs you, and the honest test for whether you need one at all.

Read guide →8 min Gateways

Portkey alternatives: what to move to, sorted by the job

Portkey does four jobs at once, so there is no single replacement. Pick by which of the four you actually needed.

Read guide →7 min
02

Every guide in this cluster

What is an LLM gateway? One endpoint in front of every model One endpoint, many providers. What a gateway actually does, what it costs you, and the honest test for whether you need one at all. 8 min The best LLM gateways in 2026, scored on what they actually do Nine gateways, seven columns, no vendor got to write its own row. Including the honest case for not running one. 7 min LLM routing: how a router decides which model answers Four things people mean by routing, how each decides, and the honest answer to whether a router saves you money. 6 min Claude Code Router: run Claude Code on other models The 36,700-star tool, what it is genuinely good at, what breaks, and the three maintained ways to do the same thing. 6 min LLM failover: surviving a provider outage without losing the turn Which errors deserve a retry, which deserve a different provider, and why a mid-task agent is the hard case. 7 min Self-hosted LLM gateway: running LiteLLM, Portkey, or Kong A gateway you can stand up in ten minutes, and the operational bill that arrives three months later. 5 min OpenAI-compatible API: what it means and what breaks The phrase promises a surface, not a behaviour. Here is the surface, and the four places compatibility quietly ends. 6 min LLM gateway pricing compared: what each one really charges Every published number in one table, plus the three volumes where the ranking flips. 5 min What is OpenRouter? One key, 400+ models, and what it costs you One API key in front of every model vendor. What it costs, what it hides, and the two cases where going direct is better. 8 min OpenRouter pricing: the 5.5% fee, free models, and what routing costs The exact fee math, a $100 worked example, free-model quotas, and a direct-versus-OpenRouter table for five models people actually use. 5 min OpenRouter alternatives: eight options, and which job each one wins Every serious OpenRouter alternative, what each actually charges, and a decision table that starts from what you are trying to fix. 6 min OpenRouter vs LiteLLM: hosted aggregator or self-hosted proxy One is a service you buy, one is software you run. The fee difference is the least interesting part of the comparison. 6 min What is LiteLLM? The SDK, the proxy, and what running it costs you The most-deployed open-source LLM gateway, what it is genuinely great at, and the upgrade treadmill nobody mentions in the README. 7 min LiteLLM alternatives: seven options, sorted by the job you need done Nobody replaces LiteLLM with a like-for-like. They replace it with whichever half of it they were actually using. 5 min Requesty review: the 5% AI gateway, tested against its own claims A hosted router with the cleanest pricing in the category, a genuine EU story, and three different model counts on its own website. 7 min Helicone review: observability first, gateway second The best answer to "what did this cost per user", with a gateway bolted on whose open-source repo has not shipped since July 2025. 6 min How to use OpenRouter in Claude Code: the working setup, and what breaks Five environment variables, one of which must be empty rather than unset, and three failure modes the docs bury. 6 min Vercel AI Gateway: what it is and when it earns its place One API key in front of hundreds of models, billed at provider list price. What it does well, and the three places it bites. 7 min Vercel AI Gateway pricing: what zero markup does and does not cover The token price is the provider's. Everything that costs extra is on a separate meter, and this is the list. 6 min Cloudflare AI Gateway: what the proxy does and what it does not A caching, logging, rate-limiting proxy in front of the provider APIs you already use. Free at the core, with two meters that are not. 6 min Databricks AI Gateway: Mosaic AI Gateway is now Unity AI Gateway The enterprise one. Renamed, moved onto Unity Catalog, and now governing agents and MCP servers as well as model endpoints. 7 min Kong AI Gateway: the AI plugins on top of the gateway you run Not a new product: a plugin family on Kong Gateway. Right if you already run Kong, expensive if you do not. 6 min Portkey review: the gateway is now Prisma AIRS AI Gateway A strong gateway with the best guardrail story in the category, now inside a security vendor. What that changes. 6 min Portkey alternatives: what to move to, sorted by the job Portkey does four jobs at once, so there is no single replacement. Pick by which of the four you actually needed. 7 min TrueFoundry review: the AI gateway, and why Bifrost is not it An enterprise LLMOps platform with a gateway inside it. Strong on Kubernetes and air-gapped deploys, priced by request count. 6 min
03

Other clusters

PricingUsage trackingInstallReferenceMCPCodexGeminiCursorWorktreesGrok BotHead to headConceptsTroubleshootingSpendInference
Try it

Every agent.
One workbench.

Continuum runs Claude Code, Codex, Cursor, Gemini, and more under the subscriptions you already pay for, with live quota gauges and spend by repo.

Download for Mac Compare tools

free app · your subscriptions · local-first