gemini-4IMAGE
Gemini 4 Argon's AA score and safety numbers, per one post
A post says Gemini 4 Argon scores 53 on AA, 23 points above Gemini 3.1 Pro, ranks first on AutomationBench-AA at 77.5%, has a 15% hallucination rate and a 0.7% prompt injection success rate.
Bee
@thtbee_Archived 28 days agoOriginally posted 2026.10.01
Bee @thtbee_Originally posted 2026.10.01
Gemini 4 Argon.. (i'd be really glad to test it) > Score: 53 on AA a huge 12 point jump, same as GPT-6 Astra (max) and 1 point ahead of GPT-6.1 Sol (max, 52) > 23 points above Gemini 3.1 Pro (30) > #1 on AutomationBench-AA at 77.5%, 6 points ahead of Claude Sonnet 5.5 (71.3%) > Hallucination rate of just 15%, vs 51% for GPT-6 Astra and 54% for GPT-6.1 Sol > $1.99 per task at the current 50% discount, 60% of GPT-6 Astra ($3.26) And look at this Prompt injection attack success rate just 0.7% (lower is better)
x.com/thtbee_/status/2105407225552568590Archived 2026.10.01 09:59