Standard benchmarks fall short for multi-turn AI agents. A new approach is the 'job interview eval,' where an agent is given an underspecified problem. It is then graded not just on the solution, but on its ability to ask clarifying questions and handle changing requirements, mimicking how a human developer is evaluated.

Progress in complex, long-running agentic tasks is better measured by tokens consumed rather than raw time. Improving token efficiency, as seen from GPT-5 to 5.1, directly enables more tool calls and actions within a feasible operational budget, unlocking greater capabilities.

AI struggles with long-horizon tasks not just due to technical limits, but because we lack good ways to measure performance. Once effective evaluations (evals) for these capabilities exist, researchers can rapidly optimize models against them, accelerating progress significantly.

Classifying a model as "reasoning" based on a chain-of-thought step is no longer useful. With massive differences in token efficiency, a so-called "reasoning" model can be faster and cheaper than a "non-reasoning" one for a given task. The focus is shifting to a continuous spectrum of capability versus overall cost.

A paradox exists where the cost for a fixed level of AI capability (e.g., GPT-4 level) has dropped 100-1000x. However, overall enterprise spend is increasing because applications now use frontier models with massive contexts and multi-step agentic workflows, creating huge multipliers on token usage that drive up total costs.

Traditional AI benchmarks are seen as increasingly incremental and less interesting. The new frontier for evaluating a model's true capability lies in applied, complex tasks that mimic real-world interaction, such as building in Minecraft (MC Bench) or managing a simulated business (VendingBench), which are more revealing of raw intelligence.

Even for complex, multi-hour tasks requiring millions of tokens, current AI agents are at least an order of magnitude cheaper than paying a human with relevant expertise. This significant cost advantage suggests that economic viability will not be a near-term bottleneck for deploying AI on increasingly sophisticated tasks.

As reinforcement learning (RL) techniques mature, the core challenge shifts from the algorithm to the problem definition. The competitive moat for AI companies will be their ability to create high-fidelity environments and benchmarks that accurately represent complex, real-world tasks, effectively teaching the AI what matters.

OpenAI identifies agent evaluation as a key challenge. While they can currently grade an entire task's trace, the real difficulty lies in evaluating and optimizing the individual steps within a long, complex agentic workflow. This is a work-in-progress area critical for building reliable, production-grade agents.