Brothers, don't say Chinese models are no good anymore.


Now the entire AI tech stack of foreigners has been directly replaced with Chinese models.
Costs dropped by 87%, revenue didn't drop.
Inference backend: Opus 4.8 replaced by Kimi K2.7, 8% difference, 11x cheaper.
Code: GPT-5.5 replaced by Qwen 3.7 Max, 18% difference, 7x cheaper.
Agent tool calling: Sonnet 4.7 replaced by GLM 5.2, 3% difference, 5x cheaper on input.
Batch tasks: GPT-5.5 mini replaced by MiMo V2.5, 6% difference, 12x cheaper.
Image: GPT-Image-2 replaced by Wan 2.5, 5% difference, 8x cheaper.
Video: Sora 2 replaced by Kling 3.0, similar quality, 6x cheaper.
Ran for 30 days:
Cost -87%, quality only dropped 4% on average, revenue unchanged.
Even crazier, it runs locally, data never leaves, and you don't have to worry about getting blocked every day.
This isn't a migration.
This is doing the math crystal clear.
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned