Bifrost

Bifrost is an AI gateway written in Go and maintained by Maxim AI, whose legal entity is H3 Labs Inc. It sits between applications and model providers and offers one OpenAI-compatible API across more than 20 providers, including OpenAI, Anthropic, AWS Bedrock, Google Vertex, Azure, Mistral, Groq and Ollama. Its features include automatic fallbacks, load balancing across keys, simple and semantic caching, virtual keys with hierarchical budgets and rate limits, an MCP gateway, custom plugins and OpenTelemetry metrics. Maxim's benchmark claims about 11 microseconds of added overhead at 5,000 requests per second on a t3.xlarge instance, but this figure leaves out provider latency. Bifrost suits developer and platform teams that want to self-host a fast gateway and can switch existing SDKs to it by changing the base URL. There are two plans. The OSS edition is free under Apache 2.0 and runs through npx, Docker, Kubernetes or a Go binary. The Enterprise edition has custom pricing and a 14-day trial. It adds guardrails, clustering, adaptive load balancing, SAML and OIDC single sign-on, RBAC, audit logs, vault integration, log exports and in-VPC or air-gapped deployment. It works with LangChain, LangGraph, the Vercel AI SDK and Maxim's separate evaluation platform. Its main limitation is that many governance controls security teams expect, such as SSO, RBAC and audit trails, are only in the unpublished-price Enterprise tier. Self-hosters also pay for and run the infrastructure, the database and on-call cover themselves.