bolt Free & Open Source AI Gateway

Never Stop Coding

AI Gateway for Multi-Provider LLMs

160+ Providers Smart Fallback Semantic Cache MCP Server Deploy Anywhere
rocket_launch Start Free Now code View on GitHub
terminal — omniroute setup
$ npm install -g omniroute
✓ OmniRoute installed globally
$ omniroute
🎉 Dashboard opens → Connect providers → Code!
npm Docker Hub License

🤖 Free AI Provider for your favorite coding agents

Connect any AI-powered IDE or CLI tool through OmniRoute — free API gateway for unlimited coding.

Claude Code
Claude Code
⭐ 67K+
 Codex CLI
Codex CLI
⭐ 61K+
 Gemini CLI
Gemini CLI
⭐ 95K+
 Cline
Cline
⭐ 36K+
 Cursor
Cursor
⭐ Editor
OpenCode
OpenCode
⭐ 10K+
 Kilo Code
Kilo Code
⭐ 15K+
 Roo Code
Roo Code
⭐ 26K+
 Continue
Continue
⭐ 24K+
 Factory Droid
Factory Droid
⭐ Tool

📡 All agents connect via localhost:20128/v1 or cloud.omniroute.online/v1 — one config, unlimited models and quota

v3.7

By the Numbers

160+
AI Providers
29
MCP Tools
13
Routing Strategies
1200+
Tests Passing
40+
Languages
4
Protocols (MCP/A2A/ACP/Responses)

What's New in v3.7

See More

160+ providers, Memory System, Skills Framework, Vision Bridge, Responses API, 29 MCP tools, and much more.

psychology

Memory System v3.5

Persistent conversational memory across sessions. Extraction, injection, retrieval, and summarization modules keep your AI context-aware between conversations.

extension

Skills Framework v3.5

Extensible skill registry with built-in and custom skills. Sandboxed execution, interception/injection pipeline, and 4 MCP skill tools for agent control.

visibility

Vision Bridge v3.6

Cross-provider image and vision support. Automatically translates image inputs between OpenAI, Anthropic, and Gemini formats. 51 test scenarios validated.

api

Responses API v3.4

Full OpenAI Responses API support. Requests are internally translated to Chat Completions format, dispatched, and response-streamed back in Responses API events.

cookie

Web/Cookie Providers v3.6

Use your ChatGPT Plus, Grok, Perplexity Pro, Blackbox, and Meta AI subscriptions as API providers. Just paste your session cookie — no API key needed.

travel_explore

10 Search Providers v3.7

Perplexity, Serper, Brave, Exa, Tavily, Google PSE, Linkup, SearchAPI, You.com, and SearXNG. Ground AI responses with real-time web data.

build

29 MCP Tools v3.7

22 core tools + 3 memory tools + 4 skill tools. Budget guards, route simulation, session snapshots, DB health, pricing sync, and agent-assisted memory/skill management.

sync_alt

Context Handoff v3.6

Cross-session context relay strategy. Transfer conversation context between providers seamlessly. 13 routing strategies including context-relay and context-optimized.

How OmniRoute Works

Install once, connect providers, and code non-stop with automatic 4-tier fallback routing.

terminal
Claude Code
code
Codex CLI
smart_toy
Gemini CLI
edit_note
Cursor / Cline
apps
Any CLI Tool
hub OmniRoute
Tier 1 · Subscription
workspace_premium
Claude, Codex, Gemini
Tier 2 · API Key
vpn_key
OpenAI, Groq, DeepSeek
Tier 3 · Cheap
savings
GLM $0.6/1M, MiniMax
Tier 4 · Free
star
iFlow, Kiro, Amazon Q

Get Started in 60 Seconds

Install globally, connect your providers, and start coding with smart auto-fallback routing.

1

Install Globally

Run one command to install OmniRoute globally on your system.

$ npm install -g omniroute
2

Connect Providers

Open Dashboard and add your API keys or OAuth connections. Free providers available!

Dashboard → Providers → Connect
3

Point Your CLI

Configure Claude Code, Cursor, Cline, or any OpenAI-compatible tool.

http://localhost:20128/v1
🐳

Prefer Docker?

Run OmniRoute as a container with persistent data volume.

$ docker run -d -p 20128:20128 -v omniroute-data:/app/data diegosouzapw/omniroute:latest

Why Choose OmniRoute?

See how OmniRoute compares to alternatives.

Feature OmniRoute LiteLLM
Providers Supported 160+ 200+
Free Tier Routing ✓ ✗
Dashboard UI ✓ ✗
Semantic Cache ✓ ✗
Circuit Breaker ✓ ✓
13 Routing Strategies ✓ ✗
LLM Evaluations ✓ ✗
Translator Playground ✓ ✗
CLI Tools Manager ✓ ✗
Custom Combos ✓ ✗
MCP Server (29 tools) ✓ ✗
A2A Protocol (Agent-to-Agent) ✓ ✗
Desktop App ✓ ✗
Usage Analytics ✓ ✓
Cost Management ✓ ✓
Docker Deploy ✓ ✓
Media Playground (Image/Video/Audio/TTS) ✓ ✗
Registered Keys API ✓ ✗
Auto-Combo Engine (Self-Healing) ✓ ✗
Web Search Providers (5) ✓ ✗
Per-Model Combo Routing ✓ ✗
160+ Provider Icons (SVG) ✓ ✗
Memory System (Persistent) ✓ ✗
Skills Framework (Extensible) ✓ ✗
Vision Bridge (Cross-Provider) ✓ ✓
Responses API ✓ ✓
Web/Cookie Providers (5) ✓ ✗
10 Search Providers ✓ ✗
Context Handoff / Relay ✓ ✗
Self-hosted & Free ✓ ✓

160+ Providers Ready

Connect via OAuth, API Key, Web Cookie, or use completely free providers.

Free (Unlimited or High Quota)
OAuth / Subscription
API Key
Web / Cookie
Self-Hosted
iFlow
iFlow AI
Qwen
Qwen Code
Kiro
Kiro AI
Gemini CLI
Gemini CLI
Claude Code
Claude Code
OpenAI
OpenAI
Anthropic
Anthropic
Google AI
Google AI
Antigravity
Antigravity
OpenClaw
OpenClaw
Groq
Groq
DeepSeek
DeepSeek
xAI
xAI (Grok)
Mistral
Mistral
Together AI
Together AI
Fireworks
Fireworks
Perplexity
Perplexity
Cerebras
Cerebras
Cohere
Cohere
OpenRouter
OpenRouter
GLM
GLM (ZhipuAI)
MiniMax
MiniMax
Moonshot
Moonshot
Nebius
Nebius
NVIDIA
NVIDIA
SiliconFlow
SiliconFlow
Sambanova
Sambanova
Novita
Novita AI
Chutes
Chutes AI
Kluster
Kluster AI
InfiniAI
InfiniAI
Targon
Targon
AI21
AI21 Labs
Lambda
Lambda
Lepton
Lepton AI
Deepgram
Deepgram
AssemblyAI
AssemblyAI
NanoBanana
NanoBanana
HuggingFace
HuggingFace
Vertex AI
Vertex AI
Alibaba
Alibaba DashScope
LongCat
LongCat AI
Pollinations
Pollinations
Cloudflare AI
Cloudflare AI
Scaleway
Scaleway
AI/ML API
AI/ML API
Puter AI
Puter AI
OpenCode Zen
OpenCode Zen
OpenCode Go
OpenCode Go
Kimi Coding
Kimi Coding
Alibaba Coding
Alibaba Coding
ElevenLabs
ElevenLabs
Cartesia
Cartesia
PlayHT
PlayHT
Ollama Cloud
Ollama Cloud
Amazon Q
Amazon Q
GitLab Duo
GitLab Duo
Cline OAuth
Cline (OAuth)
Kimi Coding
Kimi Coding
ChatGPT Web
ChatGPT Web
Grok Web
Grok Web
Perplexity Web
Perplexity Web
Blackbox Web
Blackbox Web
CrofAI
CrofAI
Azure OpenAI
Azure OpenAI
Bedrock
Amazon Bedrock
Vertex Partner
Vertex AI Partners
DeepInfra
DeepInfra
Meta Llama
Meta Llama API
Databricks
Databricks
Snowflake
Snowflake Cortex
Venice
Venice.ai
Poe
Poe
Heroku AI
Heroku AI
IBM watsonx
IBM watsonx
Runway
Runway
Docker Model Runner
Docker Model Runner
ComfyUI
ComfyUI
LM Studio
LM Studio
vLLM
vLLM
Brave Search
Brave Search
Exa Search
Exa Search
Tavily
Tavily Search
Serper
Serper
…and 100+ more including Morph, Featherless AI, FriendliAI, NanoGPT, Predibase, Bytez, Reka, NLP Cloud, Clarifai, DataRobot, GigaChat, and OpenAI/Anthropic-compatible custom providers.

Powerful Features

Everything you need to route, monitor, and optimize your AI usage.

route Routing & Reliability
swap_vert

Smart 4-Tier Fallback

Subscription → API Key → Cheap → Free. Automatic switching when quota runs out, zero downtime.

account_tree

Intra-Family Model Fallback New

When a model is unavailable, automatically falls back to sibling models in the same family before returning an error.

balance

13 Routing Strategies + Auto-Combo v3.7

Priority, weighted, round-robin, context-relay, fill-first, P2C, random, least-used, cost-optimized, strict-random, auto-combo, LKGP, and context-optimized. Per-combo or global.

offline_bolt

Circuit Breaker

Auto-open and close per-provider with configurable cooldowns. Self-healing after failures.

shield

Anti-Thundering Herd

Mutex + automatic rate-limiting for API key providers. Prevents quota exhaustion spikes.

fingerprint

Request Idempotency

5-second dedup window for duplicate requests. Saves tokens and prevents double-sends.

psychology Intelligence
cached

Semantic Cache New

Two-tier cache (exact + semantic similarity) reduces cost and latency for repeated queries.

translate

Format Translation

Seamless OpenAI ↔ Claude ↔ Gemini format translation. Use any model with any client.

psychology_alt

Think Tag Parsing

Automatically parse and handle <think> tags from reasoning models like DeepSeek R1.

security

Prompt Injection Guard New

Built-in protection against prompt injection attacks on your AI endpoints.

route

Task-Aware Smart Routing New

Automatically selects the best model based on content type — coding, analysis, vision, summarization. 7 task types.

monitoring Monitoring & Analytics
timer

Real-Time Quota Tracking

Live token consumption, reset countdown, and cost estimation per provider.

analytics

Usage Analytics New

Full dashboard with tokens, costs, trends over time. Filter by provider, model, or period.

payments

Costs & Budget New

Track spending with editable per-model pricing. Set budget alerts and limits.

health_and_safety

Health Monitor New

Dashboard with healthcheck per provider, token validation, and auto-refresh status.

science

LLM Evaluations New

Golden set testing with 4 match strategies: exact, contains, regex, custom JS function.

code Developer Experience
chat

Translator Playground New

Built-in Chat Tester and Test Bench. Test any model in real-time from the dashboard.

build

CLI Tools Manager New

Configure Claude Code, Codex, OpenClaw, Kilo, Droid, and Cline directly from the dashboard.

tune

Custom Combos New

Create unlimited model combinations with 6 balancing strategies. Fine-tune routing per combo.

group

Multi-Account Support

Add multiple accounts per provider. Round-robin load balancing and automatic failover.

image

Media Playground v3

Full media generation: Image (NanoBanana, SD WebUI, ComfyUI), Video, Music, Audio Transcription (2GB, Deepgram, AssemblyAI), and Text-to-Speech (ElevenLabs, Cartesia, PlayHT).

cloud_sync

Cloud Sync

Sync config across devices via Cloudflare Workers. 300+ global edge locations.

key

API Key Access Controls New

Create scoped API keys with model restrictions, time-based access schedules, and enable/disable toggles.

folder_open

Connection Groups New

Organize provider connections by environment (dev/prod). Accordion view with smart auto-switch.

hub Protocol & Integration
extension

MCP Server (29 Tools) v3.7

Model Context Protocol server with 29 agent-control tools (22 core + 3 memory + 4 skills). 3 transports: stdio, SSE, Streamable HTTP. 10 scoped auth levels.

lan

A2A v0.3 Protocol New

Agent-to-Agent orchestration with JSON-RPC 2.0, task streaming, SSE heartbeat, and smart-routing skill.

desktop_windows

Desktop App New

Native Electron app for Windows, macOS, and Linux. System tray, auto-update, offline support, single-instance lock.

price_check

External Pricing Sync New

3-tier pricing resolution synced from LiteLLM. User overrides → synced → defaults. Opt-in via settings.

memory Memory, Skills & AI Intelligence
psychology

Memory System v3.5

Persistent memory across sessions with extraction, injection, retrieval, and summarization. 3 MCP memory tools for agent-controlled recall.

extension

Skills Framework v3.5

Extensible skill registry with built-in and custom skills. Sandboxed executor, interception/injection pipeline, A2A integration, and 4 MCP skill tools.

visibility

Vision Bridge v3.6

Cross-provider image support. Translates vision inputs between OpenAI, Anthropic, and Gemini formats automatically. 51 test scenarios.

api

Responses API v3.4

Full OpenAI Responses API compatibility. TransformStream converts Chat Completions SSE chunks into Responses API event format.

cookie

Web/Cookie Providers v3.6

Use ChatGPT Plus, Grok, Perplexity Pro, Blackbox, and Meta AI subscriptions as API providers via session cookies.

policy

Compliance & MITM v3.7

Provider audit module for compliance enforcement. Optional MITM proxy with cert management, DNS handling, and target routing.

Beautiful Dashboard

Monitor everything in real-time. Manage providers, combos, analytics, and more.

OmniRoute Dashboard

Deploy Anywhere

Run locally, in a container, on a VM, or at the edge.

📦

npm

Install globally for local development

npm install -g omniroute
🐳

Docker

Container with persistent data volume

docker run -d diegosouzapw/omniroute
🖥️

VM / VPS

Deploy on Akamai, AWS, DigitalOcean

nginx → Docker → omniroute
⚡

Cloudflare Workers

Edge deployment with D1 database

wrangler deploy