Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Show HN: Metis – An agent harness pushing DeepSeek to Opus-tier coding (82%) (github.com/wholiver)
3 points by oliverhuchenrui 12 hours ago | hide | past | favorite | 2 comments
 help



In the final benchmark tests, DeepSeek V4 Flash achieved a score of 82.02%, coming close to the 83.8% score of Claude Fable 5 XHigh—effectively giving a "Flash" model the capabilities of a flagship model.

I think everyone will look for a comparison with Pi.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: