Back to No Priors: AI, Machine Learning, Tech, & Startups

Why Traditional Benchmarks Fail Modern AI Models with OpenAI Research Scientist Noam Brown

Name: Why Traditional Benchmarks Fail Modern AI Models with OpenAI Research Scientist Noam Brown
Uploaded: 2026-06-26T10:25:40.000Z
Duration: 36 min 19 s
Channel: No Priors: AI, Machine Learning, Tech, & Startups

No Priors: AI, Machine Learning, Tech, & StartupsJune 26, 202636m

Topics25

Traditional Benchmarks and Test Time Compute Limitations0:00 Motivation for the Essay on Large-Scale Test Time Compute0:31 Why Benchmarks Don't Show True Improvement2:01 Proper Evaluation Methods for Modern Models4:00 Handling Long-Running Evaluations4:31 Practical Use of Test Time Compute5:30 Benchmark Maxing and Gaming Concerns7:00 Risk of Optimizing for Benchmarks8:00 Personal Evaluation Methods8:30 Evolution of Model Capabilities in Poker Bot Creation9:30 Improvement with o1.5 in Poker Tasks11:00 Safety Evaluations and Responsible Scaling Policies11:31 Long-Horizon Task Evaluation Challenges14:02 Model Release Cycle vs Evaluation Time15:31 Latent Capabilities in Released Models17:00 The Temptation to Wait for Next Model Release19:31 Recursive Self-Improvement and Test Time Compute21:03 Test Time Compute and Benchmark Performance22:10 Research Capabilities and Limitations23:31 Intelligence Takeoff Scenarios25:31 Multi-Agent Systems and Knowledge Accumulation27:10 Frontier Competition Dynamics29:30 Practical Model Usage Recommendations31:01 Benchmark Evaluation Standards32:02 Routing Layers and Compute Allocation33:31

In a Nutshell

Traditional benchmarks understate model capabilities by not accounting for test-time compute scaling. Modern models like o1.5 show substantial gains when evaluated across compute budgets, but these improvements are invisible in standard benchmark grids that lack a tokens/cost/time axis. Safety evaluations and responsible scaling policies remain misaligned with this reality, as they fail to test what models can achieve with large inference budgets, creating evaluation gaps that grow worse with each release cycle.

AI-Generated Notes

These notes were generated by AI and may contain inaccuracies.

With GPT-3, you couldn't scale test time compute. If you gave it a budget of $10 million and asked what GPT-3 could do, it really couldn't do that much more than what you could do with $10 or $1. The current frameworks and responsible scaling policies don't really account for the amount of test time compute. They just say what's the capability of the model. The problem is we're in a world now where the capability of the model is a function of how much money you put into it. If you give it a budget of $10,000, it can do a lot more than what it can do with a budget of $10. Give it a budget of $10 million, you can do even more. At what budget should you evaluate these models? The policies that exist today don't really address that question.

The motivation was the release of o1.5 and the initial reaction was skepticism that it was a substantially better model. That only lasted for a few hours before people had time to play around with it and saw that it was actually substantially better. A lot of the skepticism came from the benchmark grid that was published. Whenever a new model is released, there's this benchmark grid where they show all these different benchmarks on the x-axis and the performance of different models on the y-axis. If you look on paper at the difference between o1.5 and o1.4 or other models, it was an improvement, but it wasn't a huge improvement. It was only a few percentage points in some benchmarks.

The reason why it doesn't show up as so much better on the benchmarks is because the benchmarks are being presented in the wrong way. They're not controlling for the amount of test time compute that is being used on that benchmark question. It turned out that o1.5 is just much more efficient with its thinking. If you run it at max settings, o1.4 is thinking for a lot longer. It takes longer to get back a response than o1.5. Once you control for the amount of thinking time, you can see that o1.5 is a substantial jump over o1.4. That's people's day-to-day experience with it.

When asked why not just have o1.5 think for as long as o1.4, the question becomes how long should they think for? The typical response is until the performance plateaus. The point at which it plateaus is actually really far out these days. It was true in GPT-3 land back in 2022. The models couldn't really think productively for that long. You could just run them until they plateau. It's not that far away. But what we're seeing today with modern models is that o1.5 and other models can think for weeks even before having performance plateau on some of these benchmarks. The point at which they plateau is simply too far out to reasonably test. We all need to actually enforce either a patience limit or a budget limit from a token perspective now, and that wasn't true a few years ago.

The proper way to evaluate the models now is to either have some kind of budget for the benchmark, whether it's tokens or cost or time or whatever, or plot the performance as a function of the amount of test time compute that's going into the model. Then it becomes much more clear how to compare the performance between these different models.

Sign in to read the full notes

Get access to AI-generated notes, topic timestamps, and more.

Why Traditional Benchmarks Fail Modern AI Models with OpenAI Research Scientist Noam Brown

Traditional Benchmarks and Test Time Compute Limitations00:00

Motivation for the Essay on Large-Scale Test Time Compute00:31

Sign in to read the full notes