Phlox-GW: Open-Source LLM Gateway Brings Enterprise Features to Self-Hosted Infrastructure
Key Takeaways
- ▸Phlox-GW commits to keeping enterprise features free forever, avoiding the 'rug pull' pattern of competitors who move critical features behind commercial licenses
- ▸Includes essential operational features: cost accounting and budgeting, department-level chargebacks, SSO/OIDC, rate limiting, guardrails, and high-availability clustering
- ▸Provides protocol translation between OpenAI and Anthropic APIs, allowing any compatible client to work with any provider through a single gateway
Summary
Phlox-GW, a new open-source LLM gateway, launches with enterprise-grade features like cost accounting, budget enforcement, SSO/OIDC support, and high-availability clustering—all permanently free and open-source. Created in response to the tendency of other open-source gateways to paywall critical operational features, Phlox-GW provides a fully self-hosted alternative for managing multi-provider LLM infrastructure across teams. The gateway supports OpenAI and Anthropic endpoints with full protocol translation, enabling users to route any compatible client through a single unified interface.
Key capabilities include user and departmental cost tracking and chargebacks, budget limits with spending warnings, rate limiting (RPM/TPM) at multiple levels, PII redaction and guardrails, auditing and observability via Prometheus and OpenTelemetry, and horizontal scaling via PostgreSQL-backed clustering. Available as a single binary for macOS, Linux, and Windows, Phlox-GW can scale from personal labs to enterprise multi-node deployments without licensing concerns.
- Designed for easy self-hosting with single-instance SQLite deployment for individuals, or multi-node PostgreSQL cluster for enterprise scale
Editorial Opinion
Phlox-GW addresses a legitimate friction point in the LLM infrastructure market—the expectation that truly useful operational features (cost controls, audit trails, high availability) will eventually move behind paywalls. By committing to permanent open-source licensing for enterprise capabilities, the project positions itself as a principled alternative for teams that want to avoid vendor lock-in. If it can sustain community momentum and keep pace with the rapid evolution of LLM APIs, it could become a reference implementation for how open-source infrastructure should be funded.
