WHAT NEW IN X10.28 WED

AI collections from X · 22 collections · 58,924 posts

minimax-m3.1IMAGE

Trivia quiz comparison: M3.1 Flash scores 60%/48%/62% vs DeepSeek 1017 near 100%

In a Sonnet-generated, web-grounded trivia quiz across four rounds, DeepSeek 1017 scored near 100% while M3.1 Flash scored roughly 60%, 48% and 62%, with a final round at 91% vs 62%.

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTexArchived 7 days agoOriginally posted 2026.10.20
Trivia quiz comparison: M3.1 Flash scores 60%/48%/62% vs DeepSeek 1017 near 100% (image 1 of 4)
Trivia quiz comparison: M3.1 Flash scores 60%/48%/62% vs DeepSeek 1017 near 100% (image 2 of 4)
Trivia quiz comparison: M3.1 Flash scores 60%/48%/62% vs DeepSeek 1017 near 100% (image 3 of 4)
Trivia quiz comparison: M3.1 Flash scores 60%/48%/62% vs DeepSeek 1017 near 100% (image 4 of 4)
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTexOriginally posted 2026.10.20
Open original on X

Sonnet made a trivia quiz. Both 1017 and M3.1-Flash aced the 1st batch. I told Sonnet to try harder, ground keys in web search. 2nd round: M3.1 scored ≈60%; Whale 100%. On the 3rd round, *Sonnet's key* had an error; 100% Whale vs 48% M3.1. Tier 4: 91% vs 62% T4.5 processing.

This post quotes another post: x.com/teortaxestex/status/2112413994904678549

x.com/teortaxestex/status/2112467148413227286Archived 2026.10.20 20:23