Skip to content
How it Works Providers Install GitHub
Open Source · Local First · Zero Telemetry

Stop Overpaying for AI.
Start Routing Smart.

One proxy. Every provider. Zero config pain. CORVYN routes your AI requests to the cheapest model that can handle the job — saving you up to 85% on API costs.

85% Cost Savings
15+ Providers
<0.1ms Routing Speed
Star on GitHub Read the Docs
corvyn — zsh
$ corvyn start

CORVYN USERS SAVED

₹ 0

example monthly savings with CORVYN

100% on Local Benchmark (250 tests)
Zero Telemetry — Nothing Leaves Your Machine
60+ Currencies Supported
15 Providers · 40+ Models
<0.1ms Task Classification
MIT Licensed · Open Source
100% on Local Benchmark (250 tests)
Zero Telemetry — Nothing Leaves Your Machine
60+ Currencies Supported
15 Providers · 40+ Models
<0.1ms Task Classification
MIT Licensed · Open Source

Audience

Who Is CORVYN For?

Save on AI costs no matter your size or use case.

Indie Developers

Build projects with free tiers. Handle 90% of tasks with Gemini, Mistral, and OpenRouter. Keep costs near zero while maintaining quality.

Startups

Scale from zero to millions of users intelligently. Route simple tasks to free tiers, hard tasks to your paid keys. Never surprise bills.

Enterprise

Centralize AI routing with policy controls. Track costs in your local currency. Maintain security while optimizing spending across teams.

Workflow

How It Works

Three steps to cut your AI bill by 85% without changing how you work.

You Use Any AI Tool

OpenCode, Claude Code, Cursor, Aider — just change one URL setting and you're routing through CORVYN.

CORVYN Routes Smartly

Classifies every task in <0.1ms. Routes simple tasks to free models. Hard tasks to your paid key.

You Save Money

Free models handle 85% of tasks. See exactly what you spend in your currency. Never a surprise bill.

Integrations

Every Free Model. Every Gateway. One Proxy.

15 providers. 40+ models. One config file. Enable only what you have.

Gemini
1,500 req/day
FREE
Mistral
500K tok/month
FREE
OpenRouter
Free models + Pay-as-you-go
GATEWAY
OpenCode Zen
Free + Pay-as-you-go
GATEWAY
OpenCode Go
$10/mo · 14 models
GATEWAY
Ollama
Unlimited · Local
LOCAL
Groq
Pay-per-token
PAID
Cerebras
Pay-per-token
PAID
SambaNova
Pay-per-token
PAID
Anthropic
Claude Sonnet
PAID
OpenAI
GPT-4o
PAID
DeepSeek
Cheapest paid
PAID
Kimi
K2.6 · Agentic
PAID
Cloudflare AI Gateway
20+ providers · Edge caching · Analytics
GATEWAY

Cloudflare Integration

Cloudflare AI Gateway Built In

Route through Cloudflare's edge network. Get caching, analytics, rate limiting, guardrails, and DLP — with zero extra code.

3

Billing Modes

20+

Upstream Providers

Edge

Cache & Analytics

Per-Model

Cost Tracking

# corvyn.config.toml
[providers.cloudflare_ai]
enabled = true
account_id = "$CF_ACCOUNT_ID"
gateway_id = "default"
mode = "unified"
api_token = "$CF_AIG_API_TOKEN"
models = ["workers-ai/@cf/moonshotai/kimi-k2.6", ...]

Global by Default

₹ 2,840 saved today

Auto-detected from your system locale. 60+ currencies supported with real-time exchange rates.

Quick Start

Up in 5 Minutes

Install, configure, and start routing. Works with any OpenAI-compatible client.

1Install 
curl -fsSL https://raw.githubusercontent.com/corvyn-ai/corvyn/main/install.sh | bash
2Configure 
corvyn init
3Start 
corvyn start
4Connect OpenCode 
{
  "provider": {
    "corvyn": {
      "npm": "@ai-sdk/openai-compatible",
      "options": { "baseURL": "http://localhost:4000/v1" },
      "models": {
        "auto": { "name": "CORVYN Auto Router" },
        "free": { "name": "CORVYN Free Only" },
        "paid": { "name": "CORVYN Paid Only" }
      }
    }
  },
  "model": "corvyn/auto"
}
github.com/corvyn-ai/corvyn

Built in Tamil Nadu. For developers everywhere.

MIT Licensed

Fork it, use it, build on it

Local First

Nothing leaves your machine

Community Built

PRs and issues welcome