Anthropic releases Opus 5 with efficiency gains at Opus 4.8 pricing
Anthropic launched Opus 5, saying it improves coding, reasoning efficiency, and prompt-cache-friendly tool use while keeping pricing at $5 per 1M input tokens and $25 per 1M output tokens, matching Opus 4.8. The company positions the model to deliver similar quality with fewer tokens, targeting lower cost for production agent workloads.
Key takeaway
The release frames the next competitive edge less as raw peak scores and more as cheaper, cache-aware agent runs at the same list price as the prior Opus tier.
Context
Opus 5 arrives as another top-tier Anthropic model update aimed at builders shipping multi-step tool agents, not just chat demos. Reported focus areas are coding quality, more efficient reasoning, and tool use that plays better with prompt caching.
Holding the Opus 4.8 price points means any real savings come from using fewer tokens per task rather than a sticker-price cut. That matters for teams already running high-volume agent loops where cache hit rates and token burn dominate unit economics.
Numbers to know
- $5Price per 1M input tokens for Opus 5, same as Opus 4.8
- $25Price per 1M output tokens for Opus 5, same as Opus 4.8