DeepSeek V4.1 Flash weights out, outperforming Opus-5 and GPT-5.6-Sol in key areas

Sentdex · x · 2026-09-11

The quoted post notes that DeepSeek V4.1 Flash weights are out, considerably bigger than its predecessor, outperforming Opus-5, GPT-5.6-Sol, Kimi-K3 and GLM-5.3 in several key areas.

Poster Sentdex, the well-known ML YouTuber, laments that one of his RTX Pro 6000 cards died just as the release dropped, while conceding the model "looks very good."

Related event: DeepSeek V4.1 Flash Leak: 552B Asymmetric MoE Reportedly Rivals GPT-5.6(20 posts)→

Original post →

More from Models

Models channel →