On the new VIALS-AA benchmark for scientific visual reasoning, Gemini 4 Carbon leads at 64%, ahead of GPT-6 Astra (63%) and Claude Opus 5.5 (61%), with cost-per-task also highlighted in coverage of the Artificial Analysis launch.
Need help?
Contact usOn the new VIALS-AA benchmark for scientific visual reasoning, Gemini 4 Carbon leads at 64%, ahead of GPT-6 Astra (63%) and Claude Opus 5.5 (61%), with cost-per-task also highlighted in coverage of the Artificial Analysis launch.