“I believe you will see a chatGPT level language model (on at least some metrics) on a mobile phone next year and GPT-4 level year after. These models are barely optimised, you would be very surprised.”

Emad MostaqueCEO

Founder and CEO of Stability AI

By the end of 2025 small models that run on flagship phones — e.g. Qwen3-4B-2507, Gemma 3n and Phi-4-mini — matched or exceeded the original March-2023 GPT-4 on many standard benchmarks (reasoning, math, coding), though they still lag GPT-4 in breadth of world knowledge, so the claim holds on the 'at least some metrics' standard Mostaque set in the same tweet.