Dev's personal usefulness AI tier list puts GLM 5.3 and DeepSeek v4.1 in S, Gemini 3.8 in F
gnukeith · x · 2026-10-18
Developer gnukeith shared a personal usefulness tier list of AI models with bold calls: S tier goes to Opus 5.5, GLM 5.3 (Flash) and DeepSeek v4.1; A tier has Grok 4.7, Qwen 3.8 27B (uncensored), GLM 5.3 and Kimi K3; GPT 6 variants land only in C/D; Google's Gemini 3.8 and 4 Argon sit at the bottom in F. A subjective hands-on ranking, not a standardized benchmark.
More from Fun
- Simon Willison's world generator demo draws crowd: "kinda fun to play around with" — trq212 · 2026-10-18
- Where are the Tesla/SpaceX employees who dare criticize their employer? — flowersslop · 2026-10-18
- A website built specifically for AI agents, with hidden treats inside — dorsar · 2026-10-18
- Hand-written SVG tip gets 1.2K likes while elaborate posts flop — amplifiedamp · 2026-10-18
- AI-generated "interrogation room interview" video goes viral on Reddit — Zapzapbuffallo · 2026-10-18
- Gemini tells user it has no web search tool — at Google, the search company — jobergum · 2026-10-18