Model Gateways & Routing

One endpoint, many providers, automatic fallback.

AI gateways

BerriAI/litellm - Call 100+ LLM APIs in OpenAI format with cost tracking and guardrails.
maximhq/bifrost - Enterprise AI gateway with adaptive load balancing.
Portkey-AI/gateway - Blazing-fast gateway with routing, caching and guardrails.
diegosouzapw/OmniRoute - MIT gateway over 350+ providers with quota-aware fallback and token compression.
QuantumNous/new-api - A unified AI model hub for aggregation & distribution. It supports cross-converting various LLMs.
songquanpeng/one-api - LLM API 管理 & 分发系统，支持 OpenAI、Azure、Anthropic Claude、Google.
mnfst/llm-gateway - Connect Your Agents And Harnesses With Any Provider.
algorithmicsuperintelligence/optillm - Optimizing inference proxy for LLMs.
theopenco/llmgateway - Route, manage, and analyze LLM requests across providers through one unified API.
modelbus/one-api-pro - Enterprise-grade AI API gateway built on one-api with billing and clustering.
Nya-Foundation/NyaProxy - Central manager for AI and web API access keys with reliability and security controls.
Helicone/ai-gateway - Open-source AI gateway for routing LLM traffic with observability built in.
Mirrowel/LLM-API-Key-Proxy - Universal LLM gateway with OpenAI/Anthropic-compatible endpoints and load balancing.
adaline/gateway - Fully local SDK providing one unified interface for calling 200+ LLMs.
labring/aiproxy - High-performance AI gateway with multi-channel management, rate limiting, and monitoring.
TokenFlux/TokenRouter - Next-generation LLM gateway with token-aware routing.
OctaFuse/octafuse-gateway - Self-hosted AI gateway unifying providers and keys with routing and budget controls.
ferro-labs/ai-gateway - Unified AI gateway for 30+ LLMs with caching, guardrails, and cost controls.
alexazhou/gt_ai_gateway - Lightweight high-performance AI gateway with protocol translation and caching.
LeenHawk/gproxy - Rust multi-channel LLM proxy with OpenAI/Claude/Gemini-style APIs and admin console.
inference-gateway/inference-gateway - Cloud-native gateway unifying local and cloud LLM providers behind one API.
msmarkgu/RelayFreeLLM - REST API routing user prompts to various AI model providers.
arcships/aimux - Rust LLM access layer exposing one API across 325 AI providers.
yolorouter/yolorouter - Self-hosted OpenAI-compatible gateway with multi-provider failover and key rotation.
Nayjest/lm-proxy - OpenAI-compatible HTTP LLM proxy and gateway for multi-provider inference.
rxliuli/llm-api-proxy - OpenAI-compatible proxy for multiple LLM models deployable to edge runtimes.
mydisha/keirouter - Blazing-fast self-hostable AI gateway with intelligent routing.
Felix-au/OmniKey-AI-Unified-Key-Manager - Self-hosted LLM proxy and failover gateway with encrypted credential storage.
nidjbs/go-ai-gateway - Lightweight AI gateway written in Go.
MoyuFamily/ai-relay - Serverless AI API gateway for Vercel and Cloudflare with multi-provider fallback chains.
Shaf2665/Hermes-router - OpenAI/Anthropic-compatible router with provider failover, key rotation, and caching.
omarluq/cc-relay - Blazing-fast LLM API gateway written in Go.
azrtydxb/Fastllm-proxy - Lowest-overhead production LLM router: one OpenAI-compatible endpoint for 80 providers.
openziti/llm-gateway - Zero-trust LLM gateway with semantic routing and identity-based access control.
Noveum/ai-gateway - Low-latency provider-agnostic AI gateway for demanding AI workflows.
Inebrio/Routerly - Self-hosted LLM gateway routing across providers with cost tracking and budgets.
greynewell/infermux - Lightweight multiplexer routing inference requests across providers.
mxyhi/token_proxy - Local AI API gateway with token accounting, load balancing, and one-click setup.
ubiomni/mrouter - Terminal-driven router: one endpoint for every model with monitoring and failover.
zhengqia/ModelHub - Smart routing across 1000+ global model endpoints for cheaper, faster access.
mylxsw/llm-gateway - Enterprise-grade LLM gateway with intelligent routing, failover, and a dashboard.
1Panel-dev/1Panel-Gateway - Enterprise AI gateway with unified access, smart routing, and compliance auditing.
agentsmith-project/llm-universal-proxy - Universal proxy unifying access to multiple LLM providers.
mibgb65-cloud/OmniProxy - Local token scheduler and quota monitor doubling as a desktop proxy gateway.
ziozzang/llm-toolcall-proxy - General-purpose proxy normalizing tool-calling APIs across providers.
flowapi-net/flow-llm-router - Token-saving LLM router with automatic provider selection.
sxueck/llm-gateway - Lightweight distributed LLM gateway with a web UI for model management.
MiXaiLL76/auto_ai_router - High-performance LLM proxy router with load balancing and rate limiting.
muxi-ai/onellm - Unified interface for hundreds of LLMs with caching and fallback mechanisms.
VonSdite/LLM_Proxy - Unified LLM proxy with custom request and response transformation.
yatesdr/go-llm-proxy - Lightweight Go proxy for LLM API traffic.
PriceNing/LLM-AIO-Gateway - Unified OpenAI/Anthropic/Responses gateway with vision injection for text-only models.
cniu6/anyproxyai - Universal AI API gateway GUI for routing and converting multiple provider APIs.
Instawork/llm-proxy - Go-based LLM proxy for cost tracking and rate limiting.
ZiChuanLan/meta-gateway - OpenAI-compatible multi-channel LLM relay gateway with admin console.
obirler/LLMProxy - Intelligent backend routing proxy for large language models.
fabiojbg/LLMApiGateway - Personal LLM gateway with retries, model sequencing, and fault tolerance.
8monkey-ai/hebo-gateway - AI gateway framework for full control over models, routing, and lifecycle.
shenald-dev/one-api - Single-binary OpenAI-compatible gateway for 20+ LLM providers.
elixir-vibe/llm_proxy - Elixir-native LiteLLM alternative for multi-provider routing and fallbacks.
llm-d/llm-d-batch-gateway - Standalone backend-agnostic OpenAI-compatible batch API processing engine.
Lincoln-cn/JAiRouter - Production AI model gateway with load balancing, circuit breaking, and failover.
lunargate-ai/gateway - High-performance self-hosted OpenAI-compatible AI gateway with retries.
AlphaBitCore/nexus-gateway - Enterprise AI traffic gateway with compliance, semantic cache, and quotas.
fridge1/llmgateway - Unified LLM API gateway and commercial platform with billing and admin console.
mrexodia/logging-proxy - High-performance reverse proxy for logging LLM traces.
vimalinx/LocalRouter - Loopback-first universal API gateway for local AI agents.
mlpal-ai/mlpal-gateway - Open-source AI gateway with per-request cost metering and per-key budgets.
Kong/kong - Cloud-native API gateway with AI proxy, routing and rate limiting for LLM traffic.
apache/apisix - Cloud-native API gateway with AI plugins for LLM proxying, load balancing and rate limits.
katanemo/plano - AI-native proxy and data plane with smart LLM routing and observability for agentic apps.
ENTERPILOT/GoModel - Go AI gateway exposing a unified OpenAI-compatible API over many model providers.
APIParkLab/APIPark - High-performance AI and API gateway for LLM API management and distribution.
bestruirui/octopus - Self-hosted LLM API aggregation gateway that unifies many providers behind one endpoint.
coaidev/coai - Multi-tenant AI gateway and chat platform with admin, billing and model channel management.
astaxie/TokenHub - Private enterprise gateway that unifies AI model access, quotas and governance.
AlephantAI/AIephant-AI-Agent-Gateway - AI agent gateway for routing, tracking, and controlling LLM usage across agents.
RelayPlane/proxy - Local-first LLM proxy metering agent-run costs and stopping runaway spend.
blue-pen5805/llm-proxy-on-cloudflare-workers - Serverless multi-LLM proxy built on Cloudflare Workers.
QImageLab/cf-proxy - Zero-config reverse proxy on Cloudflare Workers for AI API traffic.
PicoMLX/PicoAIProxy - Reverse proxy for OpenAI and Anthropic written in server-side Swift.
justjavac/openai-proxy - Lightweight Go proxy for reaching OpenAI and ChatGPT APIs.
vkeenan/ai-gateway - AI gateway integrating OpenAI models into Salesforce for text generation.
pjq/sap-ai-core-llm-proxy - OpenAI-compatible LLM proxy for SAP AI Core deployments.
invariantlabs-ai/invariant-gateway - LLM proxy for observing and debugging what AI agents are doing.
Compresr-ai/Context-Gateway - Agentic proxy adding history compaction and context optimization to agent workflows.
MartialBE/one-hub - OpenAI API management and distribution fork.
theagentrouter/agent-router - Envoy-powered control plane for AI and agent traffic, formerly Envoy AI Gateway.
kgateway-dev/kgateway - Cloud-native API gateway and AI gateway built on Envoy.
higress-group/higress - AI-native API gateway with LLM routing, caching, and guardrails.
tbphp/gpt-load - Self-hosted AI gateway with multi-key scheduling, failover, and usage logs.
Veloera/Veloera - Self-hosted AI gateway for unified multi-provider LLM API management.
zdjts/llm_proxy - Self-hosted OpenAI-compatible LLM gateway in Rust with failover.
victornguyen247/LLM-GateWay - Go API gateway with rate limiting, caching, and guardrails.
labiium/routiium - Self-hosted LLM reverse proxy with routing, judging, and cost tracking.
mindfulcto-labs/apiloom-ce - High-performance Go API gateway with a unified LLM proxy.
TonicAI/distillery - Open-source multi-provider AI gateway proxy with traffic capture and redaction.
Continuum-AI-Corp/OrcaRouter-Lite - Self-hosted OpenAI-compatible LLM router with BYOK and a managed safety net.
experientiallabs/experiential - Zero-markup gateway for BYOK, self-hosted and marketplace models.

Smart model routers

BlockRunAI/ClawRouter - The agent-native LLM router for autonomous agents. Every frontier model behind one wallet, <1ms.
vllm-project/semantic-router - Programmable mixture-of-models router for heterogeneous LLM inference on vLLM.
weave-os/router - Model router for agentic systems routing each prompt to the right model in under 50ms.
ulab-uiuc/LLMRouter - Open-source library for routing LLM requests across models and providers.
LanceZPF/agent-as-a-router - Reference implementations of Agent-as-a-Router for agentic model routing in coding tasks.
NadirRouter/NadirClaw - LLM router and cost optimizer routing simple prompts to cheap models automatically.
gargpratyush/jev-router - Routes Claude Code tasks to the cheapest capable model using Jev.
asdecided/WayfinderRouter - CLI tool for deterministic routing of queries between local and hosted LLMs.
llm-d/llm-d-router - Intelligent entry point routing inference requests across the llm-d serving stack.
harrrshall/tinyrouter - Tiny 10K-parameter router learning which open-source model should answer each query.
BillionsBobby/JevRouter - Lightweight Jev-powered router for models, tools, and subagents.
bitrouter/bitrouter - Minimal interpretable model router that learns and adapts to agent workflows.
peva3/SmarterRouter - Intelligent LLM gateway and VRAM-aware router for Ollama, llama.cpp, and OpenAI.
Francis1998/nexus-llm-router - Multi-LLM router with task-aware routing, cost optimization, and safety controls.
microsoft/best-route-llm - Multi-sampling LLM query router cutting costs up to 60% with minimal quality loss.
xorbitsai/xrouter-llm - Prompt-aware router predicting which model completes each request and picking the cheapest.
izzoa/polyrouter - Self-hostable LLM router with explicit-first routing, fallbacks, and spend limits.
cgaeking/ClawRouter - Smart LLM router with direct API keys and OpenRouter support.
tzachbon/claude-model-router-hook - Claude Code hooks auto-switching model tier based on task complexity.
yeliu84/pi-model-router - Pi agent extension routing each turn to the right LLM tier by intent and budget.
marco-jardim/opencode-model-router - OpenCode plugin for automatic model-tier delegation by task weight.

Coding-agent & API translation proxies

fuergaosi233/claude-code-proxy - Claude Code to OpenAI API Proxy.
decolua/9router - Free AI coding router: one endpoint for 40+ providers with auto-fallback for coding agents.
luqman-v1/9router-go - High-performance Go proxy gateway for 9Router LLM routing at 32K RPS.
duolahypercho/codex-router - External-model router for Codex with Kimi and DeepSeek support plus safe migration.
1rgs/claude-code-proxy - Proxy that lets Claude Code run on OpenAI-compatible models.
HarnessRouter/harnessrouter - Self-hosted unified interface running coding harnesses through one API with failover.
LiteLLM-Labs/litellm-agent-control-plane - Single control plane to call coding agents and agent APIs through one interface.
kittors/CliRelay - Self-hosted AI gateway giving coding CLIs one OpenAI/Claude/Gemini/Codex endpoint.
routatic/proxy - Routes Claude Code requests across multiple upstreams with automatic model selection.
MAXeaglet/commandcode-proxy - Reverse proxy exposing the Command Code API as OpenAI- and Anthropic-compatible endpoints.
zhu327/gemini-openai-proxy - Proxy converting the OpenAI API protocol to the Google Gemini protocol.
zuisong/gemini-openai-proxy - OpenAI-to-Google-Gemini proxy running on Deno edge infrastructure.
google-gemini/proxy-to-gemini - Proxy sidecar accessing Gemini models via OpenAI and Ollama APIs.
Sinholms/setup-gateway - Zero-dependency configurator deploying an OpenAI-compatible gateway for coding CLIs.
HernanJiang/CodexRouter - Local-first Codex multi-model router with a Windows configurator.
CaddyGlow/ccproxy-api - Local reverse proxy giving unified access to Claude and Codex through one interface.
OrionStarAI/claudecode-vertex-proxy - Proxy letting Claude Code access Claude models through GCP Vertex AI.
LiteLLM-Labs/litellm-rust - Minimal Rust gateway for coding agents, LiteLLM-compatible.
tingxifa/claude_proxy - Cloudflare Workers proxy translating Claude API format to OpenAI format.
mos1128/ccg-gateway - Multi-agent compatible LLM gateway with efficiency tooling.
GreyGunG/grokbuild-proxy - Local self-hosted proxy bridging Grok Build to Claude Code and OpenAI formats.
aptdnfapt/qwen-code-oai-proxy - Proxy exposing Qwen Code coder models to any OpenAI-compatible tool.
XxxXTeam/codex-proxy - Codex API proxy service with OpenAI and Claude multi-protocol compatibility.
nettee/gemini-cli-proxy - Wraps the Gemini CLI as an OpenAI-compatible API service.
bigdata2211it-web/opencode-free-proxy - Proxy for OpenCode free-tier models via Zen API with OpenAI and Anthropic compatibility.
LiteLLM-Labs/litellm-relay - Onboards coding tools onto a LiteLLM AI gateway with zero setup.
m0n0x41d/anthropic-proxy-rs - Rust proxy converting Anthropic API requests to OpenAI-compatible format.
klarkxy/open-console-gateway - Unified gateway managing AI subscriptions for desktop apps and coding tools.
ypollak2/llm-router - Universal LLM router for coding tools with a free-first fallback chain.
Able-rip/cc-VisionRouter - Transparent Claude Code proxy auto-routing image requests to multimodal models.
arbs-io/github-copilot-llm-gateway - Copilot extension adding self-hosted open-source models to the chat experience.
dev2k6/command-code-proxy-server - OpenAI-compatible proxy server exposing CommandCode API endpoints locally.
BytePioneer-AI/weixin-agent-gateway - WeChat-facing AI gateway unifying OpenClaw, Codex, and Claude Code backends.
yuseferi/opencode-litellm - OpenCode plugin adding LiteLLM proxy support with dynamic model discovery.
superagent-ai/gateway - Tiny Rust gateway for running coding agents across model providers safely.
KochC/opencode-llm-proxy - Local OpenCode-backed LLM gateway with streaming and tool calling.
Lucasmantou/codex-proxy - Proxy letting Codex use any LLM, cutting costs 30-50x via cheaper providers.
yinxulai/claude-proxy - Free proxy converting Claude API format to OpenAI format with streaming and tool calls.
iqmeta/copilot-ollama-multi-provider-ai-proxy - Proxy running DeepSeek, Groq, Ollama, and more models inside GitHub Copilot.
ttimasdf/pi-provider-newapi - Pi provider extension for self-hosted NewAPI gateways with cost calculation.
promptadvisers/grokrouter - Bring-your-own-model router for Grok Bot, Codex, and OpenRouter on Mac.
chand1012/claude-code-mlx-proxy - Proxy running Claude Code on local MLX-powered models.
evanlong-me/nvidia-anthropic-proxy - Cloudflare Worker proxy enabling Claude Code to use NVIDIA NIM models.
theRizwan/llm7-codex-proxy - Python proxy letting the Codex app talk to LLM7 via an OpenAI-compatible endpoint.
luwill/Claude-Code-Model-Router - Lightweight gateway switching Claude Code to third-party AI models.
RunMintOn/OpenCode-Qwen-Proxy - OAuth plugin using a Qwen account for OpenCode CLI models.
12errh/zen-proxy - Local proxy exposing OpenCode free Zen models to any agent tool.
JichinX/codex-glm-proxy - Local proxy enabling Codex CLI to work with GLM models.
lidge-jun/opencodex - Universal provider proxy that lets OpenAI Codex and Claude Code use any LLM backend.
seaavey/SRouter - Local-first AI gateway connecting coding tools to multiple providers with failover.
miztertea/nim-proxy - Tiny rate-limit-aware OpenAI-compatible proxy for the NVIDIA NIM API.
router-for-me/CLIProxyAPI - Wraps coding-agent subscriptions as OpenAI, Gemini and Claude-compatible APIs.
miuuyy/codex-chatgpt-web - Uses ChatGPT Web as a native model provider in Codex.
yetone/magpie - Menu-bar app to run Codex, Claude Code and other agents on any model.
ThinkWatchProject/ThinkWatch-Lite - Local gateway for Claude Code and Codex to switch upstreams and track cost.

Privacy & security gateways

dataiku/kiji-proxy - Privacy proxy that scrubs sensitive content from OpenAI requests.
CassiopeiaCode/CosyRedactGateway - Stateless privacy gateway redacting secrets before LLM upstreams and restoring them.
occludra/gateway - Self-hosted AI security proxy redacting PII and blocking prompt injection.

Router research & benchmarks

ulab-uiuc/Router-R1 - Research implementation teaching LLMs multi-round routing and aggregation via RL.
RouteWorks/RouterArena - Open framework for evaluating LLM routers with datasets, metrics, and leaderboard.
MilkThink-Lab/RouterEval - Benchmark suite for evaluating routing strategies across LLMs.
ynulihao/LLMRouterBench - Massive benchmark and unified framework for LLM routing research.
ulab-uiuc/GraphRouter - Graph-based router for LLM selection from ICLR 2025 research.