AI news story
I Tested All 5 Effort Levels of Claude Opus 4.7
Same $5/$25 pricing. Five effort tiers. 2.7x token cost between the cheapest and the most expensive.
Editor's take
Anthropic is now offering Claude 3 Opus with five distinct "effort" levels, allowing users to dial performance and cost up or down.
This move signifies a pragmatic approach to LLM deployment, acknowledging that not every task demands Opus's peak capabilities. By introducing tiered pricing and performance, Anthropic caters to a wider range of use cases and budgets, potentially democratizing access to advanced models beyond high-stakes, computationally intensive applications. This contrasts with the more uniform pricing models seen with models like OpenAI's GPT-4, aiming to capture more market share by offering granular control.
Future adoption will reveal whether this segmentation proves effective in balancing cost savings for simpler tasks against the premium for complex reasoning. It will be important to observe how developers and businesses integrate these tiers, and if a noticeable performance delta emerges in real-world applications beyond Anthropic's stated 2.7x token cost increase.