Astra “completely saturates” a user's spatial reasoning test, they say
The user says earlier models failed a sample question, but Astra kept getting answers right, prompting them to run the evaluation.
TLDR
A user says Astra “completely saturates” their spatial reasoning evaluation, declaring “LLM vision solved.” They describe testing Astra after earlier models had failed a sample question, then running the evaluation because Astra kept answering correctly. A separate post quoting that account calls GPT-6 Astra “AGI,” or artificial general intelligence.
Combined views
38.2K
2 Sources, first seen ago
