OpenAI models' Codeforces elo ratings over time

Hey there!

Welcome back to The Pulse, where we dive into interesting AI stories and trends backed by data, all presented through simple visuals.

> OpenAI’s o3 essentially equivalent to #175 best human competitive coder

> beats OpenAI’s Chief Scientist rating at 2665

> Sam Altman claims their internal model would currently be rank #50

> 44.2% increase in elo rating over 4 months (Sept ‘24 to Jan ‘25)

> top competitor DeepSeek R1’s elo rating estimated to be 2029

> IAP revenue from $30M (2022) to ~$1.3B (2024); downloads at almost 1.5B

> interest picked up after ChatGPT’s release

> majority revenue (45%) in 2024 from the US, followed by the UK (4%), Germany (4%), Japan (3%) & Canada (3%)

> China at <2% share of global IAP revenue

> n = 13,000 adults across 13 countries

> AI tools already incorporated in dating apps like Bumble (icebreaker, popular photo, etc.)

> Match Group (behind Hinge & Tinder) to increase AI investment

> will be launching tools soon to pick popular photos, write messages and offer dating coaching

.