AI Pulse · Frontier Labs Watch

Anthropic just released Claude Opus 5, and the pricing stayed put at $5 per million input tokens and $25 per million output tokens. That’s the same rate as the previous Opus, which tells you something about where the company thinks the market is headed.

This isn’t the absolute bleeding-edge model. Anthropic is calling it “near-flagship,” aimed squarely at coding, agents, and the kind of enterprise workflows that run all day, every day. It launched on Claude’s own API, AWS Bedrock, Google Cloud, and Microsoft Foundry, so you can plug it in wherever your stack already lives. There’s also a Fast Mode tier at $10 input and $50 output per million tokens that runs about 2.5 times faster, which matters if you’re running tight loops or customer-facing features where latency kills conversions.

The real bet

The move signals Anthropic’s view that most high-volume enterprise work doesn’t need the absolute frontier model at any cost. It needs something good enough, cheap enough, and fast enough to run in production without blowing the budget. The original report frames this as a play for the workflows that actually scale, not the demos that wow a boardroom once.

If you’re running agents that handle support tickets, generate code, or process documents in bulk, this is worth testing. The flat pricing means you can forecast costs without waiting for the next price drop, and the multi-cloud launch means you’re not locked into one vendor’s terms. It’s the kind of model that fits into an AI command centre where you’re routing different tasks to different models based on speed, cost, and capability, not just defaulting to the biggest name.