WHAT NEW IN X10.31 SAT

AI collections from X · 22 collections · 61,986 posts

Back to No.001 GPT-6 Astra in Practice: Tests, Use Cases, and FeedbackUpdated 20 hours ago
gpt-6IMAGE

Only GPT-6 Astra beats DeepSeek V4.1 Flash on a real-world design benchmark

On a real-world design benchmark, 11 closed models scored lower and cost more than DeepSeek V4.1 Flash, with only GPT-6 Astra scoring higher, which the author cites to question the future of closed models.

OpenDesign
@OpenDesignHQArchived 2026.09.14Originally posted 2026.09.09
Only GPT-6 Astra beats DeepSeek V4.1 Flash on a real-world design benchmark
OpenDesign @OpenDesignHQOriginally posted 2026.09.09
Open original on X

2) Are closed models doomed? 11 models scored lower AND cost more than DeepSeek V4.1 Flash across our real-world design benchmark. Only GPT-6 Astra scored higher.

x.com/opendesignhq/status/2097620926556955057Archived 2026.09.14 15:48
Open original on X