Claude Fable 5 and Sonnet 5 are live

One base URL. The entire Claude lineup.

Fable 5 and Sonnet 5 just landed, alongside every Opus, Sonnet and Haiku already in service. Point an Anthropic-compatible client at OpusMax and the only thing that changes is the base URL — same SDK, same request shape, same streaming.

base urlhttps://api.opusmax.pro

1M

context

Focused on

Claude Fable 5

claude-fable-5

Models

0

Max context

0M

Token window

0h

Endpoints to change

0

The lineup

Two new models at the front of the line.

Fable 5 is the most capable model available here — frontier reasoning over a million tokens of context, built for long agentic runs. Sonnet 5 gives you that same generation at Sonnet speed. Everything below them stays available on the same key.

fableNew

Claude Fable 5

Most capable. Frontier reasoning and long-horizon agentic work.

Context1M
opusNew

Claude Opus 5

New Opus flagship. Thinking on by default; knowledge current to May 2026.

Context1M
sonnetNew

Claude Sonnet 5

Frontier intelligence at Sonnet speed. The new default.

Context1M

Control

Built for whoever holds the keys.

Multi-tenant key management with real limits behind it: budgets, rates, expiry and usage tracking, plus an admin dashboard for teams and resellers.

01

Per-key token budgets

Every key carries its own 5-hour rolling window. Spend it, and the key waits for the reset — no surprise overage.

02

Isolated rate limits and expiry

Requests per minute and an expiry date are set per key, not per account. One noisy key cannot starve another.

03

Streaming passed straight through

Server-sent events are forwarded as they arrive. Token-by-token output behaves exactly as your SDK expects.

04

Prompt caching, uncharged

Cache reads and cache writes do not count against a key’s window. Only fresh input and output tokens do.

05

Drop-in Anthropic compatibility

The request and response shapes are the ones you already code against. No adapter, no SDK fork, no rewrite.

06

Works with the tools you use

Claude Code, Cursor, VS Code, Windsurf, Cline, Roo Code, the Python and TypeScript SDKs, and plain cURL.

Setup

Three steps, one of them optional.

shell

export ANTHROPIC_BASE_URL=https://api.opusmax.pro
export ANTHROPIC_AUTH_TOKEN=YOUR_API_KEY

Or run npx opusmax and let the wizard write it for you.

01

Get a key

Your admin or reseller issues one, with the budget, rate limit and expiry already set. Nothing to configure on your side.

02

Repoint the base URL

Two environment variables, or one run of the setup wizard. Your existing code is untouched.

03

Call any model in the lineup

Pass a model ID and go. Streaming, tool use, prompt caching and token counting all behave as they do upstream.

Get started

Change one line. Keep everything else.

The setup guide covers every client we support, end to end. If you already have a key, you are about sixty seconds from your first request.