Today, Anthropic unveiled Opus 5, the newest update for their popular coding and development model. While it outperforms older versions of its own Opus and competes with GPT-5.6-Sol from OpenAI in various tasks, this release does not mark a significant leap in capabilities.

The benchmarks show an iterative improvement rather than a revolutionary breakthrough. Anthropic claims Opus 5 offers similar performance to Fable at approximately half the cost, making it a competitive option for those prioritising efficiency over cutting-edge features.

However, Opus 5 does lack certain advanced training in cybersecurity tasks compared to Fable and Mythos, meaning it is 'substantially behind' on exploiting vulnerabilities. This decision by Anthropic highlights trade-offs between cost and functionality.

In summary, while Opus 5 represents a steady improvement over previous versions, it’s not the transformative step some might have hoped for. The emphasis remains on offering competitive performance at a lower price point.