From 7.3 to 54.4 in just 90 days!
🚀 Watch how DeepSeek is closing the gap on proprietary AI giants like Claude Opus. We put these models through a 14-test gauntlet to see if open source is finally ready for the crown.
DeepSeek made its own Flash model 7.5x better on DeepSWE — 7.3 to 54.4
Without adding a single parameter. Same size, same price, new post-training pass. If you want me to help automate manual work in your business with AI, setup a free call: Complete
We gave DeepSeek Harness its own browser.
Deepseek_ai Paired with ego lite, it can navigate the web using your existing brows... via Google
5 cents. That's the extra cost for DeepSeek V4 Pro to inspect and rebuild Linear's homepage with ego lite, seven minutes faste...
Via ego_agent
You have reached the end of the archive
All of deepseek