shipfeed· cluster insight
subscribe

home/cluster

Gemini 3 beats Claude on SWE-bench. 64.8% vs 61.2%.

models · evalsstarted last updated 1 source · 1 itemsimportance 8.40 / 10
Google DeepMindprimary source

No excerpt provided in the source feed. Open the article to read the full piece.

read on Google DeepMind