← Back to archive

Chinese Open-Source Models Are Surging Ahead

Looks like Chinese open-source models have surged ahead of US models.

A big driver is Qwen - fast/cheaper than most alternatives. DeepSeek is also gaining serious momentum, offering impressive reasoning at a fraction of the cost. making it an easy pick for high-volume workloads.

OpenAI is still excellent, especially for premium reasoning but a hybrid strategy works better here:

Qwen for high-volume, less-critical tasks OpenAI for premium / highest-accuracy tasks

.... meanwhile, Kimi K2 Thinking is showing very promising results. It scored 44.9% on Humanity’s Last Exam (higher than GPT-5 and every other model). Pricing is strong too: ~$0.6/M input, ~$2.5/M output.

The balance of power is moving east...

Chinese Open-Source Models Are Surging Ahead
Chinese Open-Source Models Are Surging Ahead

Originally posted on X

PS: I'm pretty active on 𝕏 if you'd like to follow along

Archive

2026

2025

2024

2018