WeirdML chart suggests DeepSeek has stalled after 3.2-Speciale
teortaxesTex · x · 2026-07-28
A chart shared with the post shows WeirdML’s interactive model comparison, and the author uses it to argue that DeepSeek has “hit a wall” after 3.2-Speciale.
The image compares multiple vendors across 17 tasks over time, with DeepSeek models clustered in the middle of the pack while the newest frontier models continue to pull ahead.
Related event: WeirdML Chart Jokes About DeepSeek Hitting a Wall(2 posts)→
More from Models
- Reddit turns Opus 5’s effort slider into a “clanker mode” meme — Fragsworth · 2026-07-28
- Grok 4.5 looks unusually token-efficient in CursorBench 3.2 — zainhas · 2026-07-28
- GPT shows a strange Spanish typo or rendering glitch in a short output — Angaisb_ · 2026-07-28
- A WeirdML chart turns DeepSeek’s progress into a running joke — teortaxesTex · 2026-07-28
- Claude screenshot jokes about “absolute truesight” and user-engineered lies — Sauers_ · 2026-07-28
- Claude Sonnet 5 gets a full meme personality review in one screenshot — Sauers_ · 2026-07-28