New benchmarks measure personal agent value and cost efficiency
SemiAnalysis and OpenAI released new benchmarks to evaluate the economic value and everyday utility of AI agents. The metrics highlight a shift toward cost-efficiency and practical tasks, with Gemini 4 Carbon offering near-parity performance at a fraction of the cost of Opus 5.5.
Two new benchmarks have been introduced to measure the value of personal agents, signaling a shift in focus toward cost-efficiency and real-world utility. SemiAnalysis released GDPX, a benchmark focused on economically valuable tasks such as chip design, finance, and biology. In this test, Opus 5.5 scored 12% at a cost of $51 per task, while Gemini 4 Carbon scored 10.9% at $16 per task. This makes Gemini 4 approximately four times cheaper with only a 1.1% lower performance score.
OpenAI simultaneously released LifeVal, designed to measure AI value in everyday life tasks, including finding childcare, accessing government services, and budgeting. OpenAI's scoring shows a steady increase in capability from GPT 5.2 to GPT 6 Astra, with success rates rising from 49.3% to 80.6%. The company stated it is now measuring and hill-climbing against these everyday tasks with free users in mind. OpenAI aims to build AI that helps with practical challenges for the more than 1 billion people who use ChatGPT weekly. The introduction of these benchmarks reflects a broader industry prioritization of making personal agents more accessible and economically viable for a broader user base, moving beyond raw capability metrics.