Per-Key Token Budgets
Set 5-hour rolling window limits per API key. Automatic resets — no manual intervention.
Full Claude Lineup — One API Key
Opus 4.7, Sonnet & HaikuReady in 60 Seconds
Get instant access to the full Claude model lineup. Drop-in Anthropic API compatible — just swap your base URL and start coding with Claude Code, Cursor, or any SDK.
Streaming SSE • Per-Key Budgets • Prompt Caching
claude-code — bash
$ export ANTHROPIC_BASE_URL=https://opusx.vercel.app/api
$ export ANTHROPIC_API_KEY=sk-ant-ox-...
$ claude
✓ Connected to OpusX Gateway
Model: claude-sonnet-4-6
Budget: 2.4M / 5M tokens
Window: 4h 12m remaining
Everything you need to manage AI access
Usage Tracking
Every request logged with model, tokens, latency, and status. Full admin dashboard.
Key Isolation
Each key gets independent rate limits, budgets, and expiry. Multi-tenant by default.
Streaming Support
Full SSE streaming pass-through. First tokens arrive instantly — zero buffering.
Drop-in Compatible
Anthropic API-compatible. Works with Claude Code, Python SDK, TypeScript SDK.
Smart Caching
Prompt caching enabled by default. Cache tokens excluded from user budgets.
Start in 3 steps
1
Get Your Key
Receive an API key from your admin or reseller.
2
Set Base URL
Set ANTHROPIC_BASE_URL to https://opusx.vercel.app/api (includes /api).
3
Start Building
Use Claude Code, SDK, or cURL exactly like Anthropic API.
Full Claude lineup
Premium
Opus 4.7
claude-opus-4-7
Most capable flagship.
Popular
Sonnet 4.6
claude-sonnet-4-6
Best balance of speed and intelligence.
Fast
Haiku 4.5
claude-haiku-4-5-20251001
Fastest responses for high throughput.