Ranking Leading AI Models from Top Labs

My Current Rankings On a Few Different Dimensions

💡 Pure Intelligence

1. Opus 4

2. o3

3. Gemini 2.5 Pro

🛠️ Real World Utility

1. o3

2. Opus 4

3. Gemini 2.5 Pro

💰 Pareto Optimality

1. Gemini 2.5 Pro

2. o3

3. Opus 4

Notes on each model (from personal experience)

Opus 4 - Launched just yesterday, May 22nd, 2025. Fair to say it will be the leading model in terms of intelligence as its the newest. But also the most expensive and still fairly rate limited. I am a Claude Pro subscriber and ran out of Opus messages yesterday after one moderately long convo.

o3 - The tool use ability and UX in ChatGPT, along with referencing chat history really makes a big difference. I’m also a ChatGPT Pro subscriber so I have access to Codex and that tool is just bonkers for producing working code.

Gemini 2.5 Pro - Very capable / cost effective model and likely the one that you might use if you’re developing an app to serve to others, where the underlying model that’s being used is not a concern for your users. Don’t have access to Gemini Deep Think yet which should rival the capabilities of o3 and Opus 4 so that very well may change the equation.