Opus 4.6 and Opus 4.7 Token Comparison

The recent release of Opus 4.7 has sparked interest in the developer community, with many wondering how the new model compares to its predecessor, Opus 4.6. Despite Anthropic's announcement that prices remain unchanged, the reality is more complex. The new tokenizer in Opus 4.7 can produce up to 35% more tokens for the same input text, potentially increasing costs.

## What happened Opus 4.7 was released on April 16, 2026, with a new tokenizer that can produce up to 35% more tokens for the same input text. This change can lead to increased costs, even though the official price remains the same. The tokenizer change is intended to improve the model's accuracy and instruction-following capabilities. According to the official release notes, the new tokenizer is a key factor in the model's improved performance. ## Why it matters The increased tokenization density in Opus 4.7 can have significant implications for developers who rely on the model for their applications. The potential cost increase can be substantial, especially for large-scale workloads. It is essential for developers to understand the pricing mechanics and run the math on their specific workloads to determine the effective cost delta. The cost increase can be mitigated by using prompt caching and batch processing, which can provide discounts of up to 90% and 50%, respectively. ## How to think about it When evaluating the costs and benefits of using Opus 4.7, developers should consider their specific workloads and use cases. It is crucial to replay real traffic side by side and measure the effective cost delta to determine the best approach. Developers should also consider the tradeoffs between the potential benefits of Opus 4.7 and the potential costs. ## FAQ