LLM Gateway

LLM Gateway is a unified API proxy and routing layer that sits between your application code and 40+ LLM providers (OpenAI, Anthropic, Google, Mistral, Cohere, and others) covering 200+ models. Instead of maintaining separate SDKs, API keys, retry logic, and cost dashboards for every provider you touch, you point your existing OpenAI-compatible client at LLM Gateway's base URL and get one endpoint that handles model selection, failover, spend tracking, and observability. It is aimed at engineering teams building production AI features who want to avoid vendor lock-in, keep a live view of token spend by model and route, and swap providers without a code change when a specific model outperforms on a given workload or when one vendor has an outage. Typical workflows include routing cheap classification traffic to smaller models while sending long-context reasoning to frontier models, A/B testing prompts across Claude and GPT variants, adding automatic fallback so a single provider outage does not take a product down, and enforcing guardrails (prompt-injection filters, PII redaction) uniformly regardless of the underlying model. The pricing structure is unusually flexible: you can bring your own provider keys and pay nothing to LLM Gateway (just for analytics and routing), buy credits that add a flat 5% margin on top of provider rates for pay-as-you-go convenience, or self-host the AGPLv3 codebase entirely for free. An enterprise tier adds SSO, dedicated support, and white-label options. Because the wire format is OpenAI-compatible, adoption is typically a one-line base-URL change, and the same integration keeps working as you add new providers behind it.