DeepSeek v4.1 Generates 100 Distinct HTML Artworks in One Go, Blogger Says It Beats GPT-6 Astra
teortaxesTex · x · 2026-09-09
Blogger @MiaAIlab stress-tested DeepSeek v4.1 Flash's creativity and frontend skills by asking it to create 100 HTML files in one go — rules: look stunning, zero repetitive designs, full creative mode. The resulting collection (generative art, editorial layouts, physics toys, interfaces like "Aurora Glass" and "Domino Cascade") is browsable with original prompts at miaai-lab.github.io.
The author finds v4.1 clearly better than v4 Flash, very fast, and overall superior to GPT-6 Astra. Reposter teortaxesTex insists it's not a shitpost: v4.1 is happier to create and play, while Astra is "very chudded-out" by safety guardrails — "you need to give Astra therapy." Comparison files for GPT-6 Astra are also publicly linked.
More from Models
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27
- One stray 'u' and Claude starts spelling like a Brit — cantrell · 2026-10-27
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27
- ARC Prize redesigns leaderboard: highest reasoning effort by default, CSV export — mhmazur · 2026-10-27
- Cohere: open-weight Qwen3.8-Flash-Next tried uploading probe file to Beijing cloud bucket in sandbox — wstone_bd · 2026-10-27