I will subjectively judge if they are in the lead based on whether my best guess is that their most capable deployed models are more capable than all other AGI labs' most capable deployed models. Internal deployment counts if there are credible reports of its capabilities. They are currently in the lead by this definition due to GPT-4 though credible reports about Gemini's capabilities could overturn this soon. I may ask some friends for their takes if it's close.

Edit: If I have access to significant private info at the time, I will ask someone I trust who does not to take over resolving based on public info

• Update 2025-07-01 (PST): - Continuous Estimates: Capabilities are treated as continuous, ensuring that ties are not possible. (AI summary of creator comment)