While AI solving long-standing math problems is impressive, its real value is questionable if those problems lack economic significance. The key indicator of AI's impact is its ability to solve problems that unlock tangible economic value, a test many current 'breakthroughs' have yet to pass.

Many influential AI model benchmarks focus on raw capabilities, like problem-solving accuracy, but neglect a critical business metric: the cost to achieve that result. Future benchmarks must incorporate the dollar cost per task to provide a more practical assessment for commercial applications.

Standardized benchmarks may not capture the true impact of a new AI model. A more effective evaluation is to apply the model to a long-standing, ambitious project that previous models failed to solve. Success on a personal "Everest" signals a genuine step-change in capability.

OpenAI's new GDPVal framework evaluates AI on real-world knowledge work. It found frontier models produce work rated equal to or better than human experts nearly 50% of the time, while being 100 times faster and cheaper. This provides a direct measure of impending economic transformation.

Leading AI models offer different trade-offs in speed, cost, and capability. A model like GPT-5.6 might be faster and more affordable for 95% of tasks, while a competitor like Fable might be superior for the most complex problems, creating a multi-leader market where different tools are used for different jobs.

Though leading closed-source models are marginally superior, open-source alternatives provide a much better price-to-performance ratio. Users pay a steep premium for the last few percentage points of intelligence offered by proprietary models, making open source a highly cost-effective choice for many applications.

When multiple models can solve a task reliably ('benchmark saturation'), the strategic goal is no longer to find the most intelligent model. Instead, it becomes an optimization problem: select the smallest, cheapest, and fastest model that still meets the performance bar, creating a major competitive advantage in inference.

Alex Karp argues that an AI's high score on a single benchmark is irrelevant for enterprise adoption. Real institutions require passing thousands of consecutive, differentiated tests. An AI model that is brilliant at one task but fails at the 50th in a complex sequence is effectively useless.

The hype around future model improvements overshadows a key reality: current models are already "sufficiently intelligent" for countless valuable tasks. Even if all AI innovation stopped today, we could still unlock trillions in economic value just by integrating existing technology across the economy.