We'll have a time-horizon of about 26 hours. We might even have labs using a serious amount of their compute to do autonomous research by that point. This might not actually show up as improved models until September-ish though (because the doing and the implementing of the research is not instant)

Assuming the doubling-time is currently around 3 months, things should be pretty crazy by June–July.

Over the subsequent months (particularly August–November) we could see Google steadily separate itself from the other labs in terms of capabilities.

But Google isn't very far behind. Its compute advantage could allow it to break away from the pack with Gemini 3.5 (July-ish), which could be a crucial moment (METR-50 time-horizons of ~50 hours, continual learning, novel insight, autonomous ML researchers).

Google seems best-positioned but is still playing catch-up. OpenAI is still in the lead. Anthropic is close behind. xAI also playing catch-up.

Google has the best trajectory, the most compute, TPUs, and the most money. OpenAI currently has the most powerful models. Anthropic has very good coding models and TPUs. xAI has Colossus II and Elon Musk.

1.4x is a lower-bound to expectations. AI will be more helpful by end-of-year than at the time the study is conducted.

I'm working on getting my head around the model of takeoff used here. I'm starting by reading this post:
www.forethought.org/research/wil...

I'm agnostic on whether models can get more performance from latent space reasoning. If so, I expect we'll see this transition instead of one to thinking in tokens of a nonhuman language, but it will still take place at the point where models get to a near-human level of understanding of the world.

I think many of the reasons that people give for discounting OpenAI in the race to AGI fall into this category. OpenAI's models have defined the curve of AI model capabilities across time so far. I expect this to continue at least in the short term.

This kind of truth is important to keep in mind when trying to divine from a lab's statements or actions how it is expecting its future models to perform.

"The other labs that are dependent on fundraising have downplayed such talk exactly because it is counterproductive for raising funds..."
From Zvi's 'AI #153: Living Documents'
www.lesswrong.com/posts/bSQagZ...

But once AI models get to a near-human level of understanding of the world, the benefits of using human languages will be outweighed by the cost of using a language optimised for a different kind of intelligence.

What I am speculating is that model CoT will move from human language to languages developed by models during training.

At the moment, it makes sense for models to use human language. It encodes a lot of useful information about the world.