Turn your $200/mo Claude Max subscription into a full OpenAI-compatible API. Stop paying per token.
Your Claude Max subscription includes unlimited* Claude usage through the CLI. This proxy wraps that CLI and exposes a standard OpenAI API, so any tool — Continue.dev, Cursor, custom apps, OpenClaw — can use your Max subscription instead of expensive API keys.
* Subject to Anthropic's fair use policy
| Approach | Monthly Cost | Notes |
|---|---|---|
| Claude API (Opus) | $15/M in + $75/M out | Adds up fast |
| Claude Max | $200/mo flat | CLI only, no third-party API |
| This Proxy | $0 extra | Uses your existing Max subscription |
Heavy users save $500-2000+/month. If you're already paying for Max, this is free money.
# Install globally
npm install -g claude-max-api-proxy
# Start the proxy (requires Claude CLI authenticated)
claude-max-api &
# Test it
curl http://localhost:3456/v1/modelsOr clone and run:
git clone https://github.com/atalovesyou/claude-max-api-proxy.git
cd claude-max-api-proxy
npm install && npm run build
npm startYour App (any OpenAI client)
|
v
Claude Max API Proxy (this) <-- localhost:3456
|
v
Claude Code CLI (subprocess)
|
v
Your Max subscription (OAuth)
|
v
Anthropic API --> Response --> OpenAI format --> Your App
Anthropic blocks OAuth tokens from direct third-party API use. But the CLI can use them. This proxy bridges that gap.
- OpenAI-compatible API — Drop-in replacement for any OpenAI client
- Streaming — Real-time token streaming via SSE
- Usage tracking — See token counts, cost savings, and request history
- API key auth — Optional Bearer token auth for team/shared use
- Multiple models — Opus, Sonnet, and Haiku
- Session management — Conversation context across requests
- Auto-start — macOS LaunchAgent for always-on service
- Secure — Uses
spawn()(no shell injection), no API keys stored
See exactly how much you're saving:
# Get usage summary
curl http://localhost:3456/v1/usage
# Response:
{
"totalRequests": 847,
"totalInputTokens": 12500000,
"totalOutputTokens": 3200000,
"estimatedApiCostSavedUsd": 427.50,
"avgResponseMs": 2340,
"byModel": {
"opus": { "requests": 523, "estimatedCostUsd": 389.20 },
"sonnet": { "requests": 324, "estimatedCostUsd": 38.30 }
},
"maxSubscriptionCostUsd": 200
}
# Recent requests
curl http://localhost:3456/v1/usage/recent?limit=10Secure your proxy for team use:
# Start with API keys
API_KEYS=sk-team-abc123,sk-team-def456 claude-max-api
# Clients must include Bearer token
curl http://localhost:3456/v1/chat/completions \
-H "Authorization: Bearer sk-team-abc123" \
-H "Content-Type: application/json" \
-d '{"model": "claude-opus-4", "messages": [{"role": "user", "content": "Hello!"}]}'When API_KEYS is not set, auth is disabled (backwards compatible).
| Endpoint | Method | Description |
|---|---|---|
/health |
GET | Health check + usage summary |
/v1/models |
GET | List available models |
/v1/chat/completions |
POST | Chat completions (streaming & non-streaming) |
/v1/usage |
GET | Usage stats and cost savings |
/v1/usage/recent |
GET | Recent request log |
| Model ID | Maps To | API Price (saved) |
|---|---|---|
claude-opus-4-6 |
Claude Opus 4.6 | $15/$75 per M tokens |
claude-opus-4 |
Claude Opus 4 | $15/$75 per M tokens |
claude-sonnet-4 |
Claude Sonnet 4 | $3/$15 per M tokens |
claude-haiku-4 |
Claude Haiku 4 | $0.25/$1.25 per M tokens |
Provider-prefixed IDs also work: anthropic/claude-opus-4-6, claude-max/claude-opus-4-6, etc.
{
"models": [{
"title": "Claude (Max)",
"provider": "openai",
"model": "claude-opus-4",
"apiBase": "http://localhost:3456/v1",
"apiKey": "not-needed"
}]
}from openai import OpenAI
client = OpenAI(
base_url="http://localhost:3456/v1",
api_key="not-needed"
)
response = client.chat.completions.create(
model="claude-opus-4",
messages=[{"role": "user", "content": "Hello!"}]
)Built-in support — just configure the claude-max provider pointing to localhost:3456.
# Non-streaming
curl -X POST http://localhost:3456/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"model": "claude-opus-4", "messages": [{"role": "user", "content": "Hello!"}]}'
# Streaming
curl -N -X POST http://localhost:3456/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"model": "claude-opus-4", "messages": [{"role": "user", "content": "Hello!"}], "stream": true}'# Create LaunchAgent
cat > ~/Library/LaunchAgents/com.claude-max-proxy.plist << 'EOF'
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>com.claude-max-proxy</string>
<key>ProgramArguments</key>
<array>
<string>/usr/local/bin/node</string>
<string>/path/to/claude-max-api-proxy/dist/server/standalone.js</string>
</array>
<key>RunAtLoad</key>
<true/>
<key>KeepAlive</key>
<true/>
</dict>
</plist>
EOF
launchctl load ~/Library/LaunchAgents/com.claude-max-proxy.plist- Claude Max subscription ($200/mo) — claude.ai
- Claude Code CLI installed and authenticated:
npm install -g @anthropic-ai/claude-code claude auth login
src/
├── adapter/ # OpenAI <-> CLI format conversion
├── server/ # Express server, routes, auth
├── session/ # Conversation session management
├── subprocess/ # Claude CLI process management
├── usage/ # Token tracking and cost analytics
└── types/ # TypeScript type definitions
spawn()instead of shell execution (no injection attacks)- No API keys stored or transmitted
- All auth handled by Claude CLI's secure keychain
- Optional API key auth for shared deployments
This proxy uses the official Claude Code CLI (claude --print) as a subprocess. It does not extract OAuth tokens, reverse-engineer private APIs, or bypass authentication — it simply wraps the CLI you already have installed.
That said, please review Anthropic's Terms of Service before using this tool. Anthropic's policies on third-party tooling may change. Use at your own discretion and risk.
PRs welcome. Please include tests.
MIT
