WHAT NEW IN X10.29 THU

AI collections from X · 22 collections · 60,637 posts

gemini-4IMAGE

Gemini 4 Argon benchmark numbers surface, including 77.9 on DeepSWE

The post lists scores such as AutomationBench 51.3, GraphWalks 1M 84, LVBench 91.7 and DeepSWE 77.9, claiming Argon beats Astra and other frontier models across the board.

Astro Polo
@astropol0Archived 28 days agoOriginally posted 2026.10.01
Gemini 4 Argon benchmark numbers surface, including 77.9 on DeepSWE (image 1 of 2)
Gemini 4 Argon benchmark numbers surface, including 77.9 on DeepSWE (image 2 of 2)
Astro Polo @astropol0Originally posted 2026.10.01
Open original on X

🚨 FINALLY GOOGLE SHOWED UP Gemini 4 Argon is landed and holy moly it mogs every Frontier in benchmarks AutomationBench 51.3 vs Astra 41 Harvey 19.6 vs single digits 1M GraphWalks 84 vs 72 LVBench 91.7 DeepSWE 77.9 Vals 68.9 Google finally cooked everyone

x.com/astropol0/status/2105392980094033991Archived 2026.10.01 10:01