5-model blind test: DeepSeek builds Adidas shoe site for $0.23 vs Opus 5.5's $33
cedric_chee · x · 2026-10-18
- The author benchmarked five frontier models on one identical task: reconstruct the Adidas Adizero Adios Pro Evo 3 running shoe in Blender (headless) from official studio shots, then build an interactive product page around the 3D model.
- Compared dimensions: cost, build time, token usage, 3D accuracy, render fidelity, and design taste. Per the results table: DeepSeek V4.1-Flash-0711 cost just $0.23 with a 73-min build; Opus 5.5 cost $33.46, took 91 minutes and burned 68.7M input tokens; GPT-6.1 Sol cost $7.84 but showed poor 3D accuracy; GPT-6 Astra was fastest (43 min, 7M tokens, $13.31); Fable 5.1 was priciest at $65.31.
- From the screenshots, DeepSeek and Opus produced shoe geometries closest to the reference photo, while GPT-6.1 Sol's model was visibly deformed; copywriting and layout styles also diverged widely. The follow-up tweet shares the full prompt.
- A useful cost/quality reference point for anyone running real engineering tasks across multiple models.
Related event: DeepSeek beats Opus in blind Blender benchmark at fraction of cost(3 posts)→
More from Models
- Ex-Microsoft exec Parakhin: after 2 days with AI advisors, I still add value — MParakhin · 2026-10-18
- $20 plan showdown: user finds Codex far less token-efficient than Claude — cneuralnetwork · 2026-10-18
- Dave Shapiro ditches ChatGPT's Dot agent over silent failures and opaque model routing — _AustinCalvert_ · 2026-10-18
- OpenAI releases 700+ AI-generated math papers, with claimed progress on Riemann hypothesis — PolarBearby · 2026-10-18
- Steering a single neuron only works for concepts with good pretraining representation — Sauers_ · 2026-10-18
- Every Grok Bot task ends with writing a prompt for local ChatGPT instead — lucasmeijer · 2026-10-18