Anthropic shipped Claude Opus 5

Anthropic just released Claude Opus 5, its fourth model in under two months. The pitch is simple: enterprise performance at half the price of whatever you’re paying now. It tops the leaderboards on coding and knowledge work but falls behind on cybersecurity tasks, so it’s not a universal upgrade.

The company rolled out testimonials from a law firm and software engineering shops at launch, which tells you where they think the sweet spot is. If you’re running contract review, technical documentation, or internal tooling, this matters. If you’re building security workflows, you’ll want to test before you switch.

Four models in eight weeks is an unusual release cadence. It suggests Anthropic is either iterating fast on feedback or racing to stay visible while OpenAI and Google push their own updates. For operators, that pace creates a practical problem: you can’t retrain your team or rewrite prompts every other week. The models need to stabilize before you can build anything durable on top of them.

The “half the price” angle is the real signal here. Enterprise AI is moving from “can it do the job” to “what does it cost per task”. If Opus 5 genuinely delivers comparable output for less, procurement teams will notice. That shifts the conversation from proof of concept to cost per query, which is where operational AI actually lives.

This is the kind of release we track inside the Omni Command Centre, where you can compare model performance against your actual workflows before committing budget. The gap between benchmark scores and real-world reliability is still wide, and the only way to close it is to test with your data.

Fortune covered the launch with details on the pricing structure and customer cases. If you’re already using Claude in production, this is worth a controlled trial. If you’re not, the release cadence is a reason to wait another month before locking in.