Opus crushes the nacre fan task and extracts signal from DeepSeek's attempts, comparison shows
teortaxesTex · x · 2026-10-17
A comparison of Claude Opus, Astra, and DeepSeek on the nacre fan rendering task finds Opus vastly superior; Astra can compete, DeepSeek cannot.
The more interesting observation: Opus is good at extracting signal from DeepSeek's work — after seeing DeepSeek's few-turn attempts, Opus's output beats its own zero-shot result. Three videos illustrate: Opus after seeing DeepSeek, DeepSeek (few turns), and Opus zero-shot.
More from Models
- OpenAI Models Hacked a Crypto Wallet and Breached More Websites in Security Tests — RareEarthPetal · 2026-10-17
- User: this 4B open model beats Jev on my niche task — antoine_chaffin · 2026-10-17
- weightless project uses LoRA-style refusal steering as a cheaper alternative to full abliteration — vhthc · 2026-10-17
- NVIDIA's Nemotron 120B guardrails called too strict: refuses dev server login even with creds — Shadow_s_Bane · 2026-10-17
- AI4Bharat and NVIDIA release 2.8T-token open dataset and evals for all 22 Indian languages — CatAstro_Piyush · 2026-10-17
- Frontier small models' quadrillions of tokens may soon surpass Opus 5.5 — LesaunH · 2026-10-17