OpenRouter alternatives, compared on where your prompts go

The most widely deployed self-hosted AI gateway, with the widest provider coverage anywhere.

• MIT except the enterprise/ directory, 57,637 GitHub stars (August 2026)
• Rust core with a Python SDK, describes itself as calling 100+ LLM APIs
• Enterprise pricing not published
• Free self-hosting of the core proxy

Best for: Teams whose priority is provider breadth and who can live inside the MIT core.

Watch out: Virtual keys, budgets, RPM and TPM limits, spend by team and SSO are enterprise features.

Maxim AI's Go gateway, built for low overhead and simple deployment.

• Apache-2.0 with no enterprise directory, 7,675 GitHub stars (August 2026)
• Docker, Kubernetes or a single Go binary
• Claims under 100 microseconds of overhead at 5,000 requests per second
• OSS tier free forever, Enterprise custom pricing

Best for: Teams where raw proxy overhead is the binding constraint, or where the OSS tier genuinely covers the requirement.

Watch out: Guardrails, Cluster Mode for high availability, adaptive load balancing, RBAC, audit logs, log export and SSO are all Enterprise. High availability being paid makes the free build closer to a development gateway.

AI plugins on top of Kong Gateway, the Apache-2.0 API gateway with years of production hardening.

• Core Kong Gateway is Apache-2.0, 44,059 GitHub stars (August 2026)
• Six AI plugins free, seventeen require an AI Gateway Enterprise licence
• Konnect Plus includes five proxied LLMs, then $100 per month per model
• Runs self-hosted, hybrid, DB-less and on Kubernetes

Best for: Teams already running Kong for their APIs, where AI traffic becomes another route rather than another system.

Watch out: Semantic caching, advanced rate limiting and every guardrail plugin are enterprise-only. Kong has also stated that Gateway 3.18 makes AI plugins opt-in rather than bundled.

A Cloudflare platform service providing analytics, caching, rate limiting and fallback.

• Not open source, no self-hosting option
• Available on all Cloudflare plans, including free
• Caching served from Cloudflare's edge rather than the model provider
• One line of code to integrate

Best for: Teams already on Cloudflare who want analytics, caching and fallback over their own provider keys, with no data residency constraint.

Watch out: Scope stops at the request boundary, and traffic traverses Cloudflare.

An open source gateway and observability platform with a very low-friction integration.

• Apache-2.0 with no enterprise directory, 6,116 GitHub stars (August 2026)
• Maintenance mode since Mintlify's acquisition announced 3 March 2026; security updates, bug fixes and new models still ship
• Hobby free with 10,000 requests, Pro $79/month, Team $799/month
• A base URL change plus one auth header

Best for: Teams where observability plus routing is the whole problem and the hosted tiers suit them.

Watch out: Both Helicone and Mintlify describe it as running in maintenance mode after the March 2026 acquisition, which constrains new feature work. Supported on-prem deployment sits under Enterprise, and SOC 2 and HIPAA compliance sit on Team at $799 per month.

A gateway that ships as part of a control plane covering the agent runtime, sandboxes, memory and evaluations.

• Apache-2.0 across the repository, with no separately licensed directory
• Virtual keys, budgets, per-tenant limits, guardrails and clustering in the free build
• Starter $0, Build $20/month, Scale $100/month (August 2026)
• Self-hosted with Docker or Kubernetes at any tier

Best for: Teams who need prompts and responses to stay on infrastructure they control, with their own provider contracts and no fee on their own keys.

Watch out: Model breadth is narrower than OpenRouter's, and you take on the provider relationships and billing OpenRouter was handling. If trying every new model the day it lands is the job, OpenRouter is better at it.