Topic
Ship in a bottle
Also searched as: ship in a bottle ai coding test
The ship in a bottle test, run by 3 models across 2 posts on X. Watch each result, then compare what the posts report below.
By model
2 postsNot a controlled comparison: prompts, tools and time differ.
Results by model
| Model | Post | Time | Cost | Tokens | Author’s verdict | Original |
|---|---|---|---|---|---|---|
| Claude Opus 5.5 | Interactive ship in a bottle@Vectorxeth | — | $12.83 | — | — | X post |
| Claude Opus 5.5 | Ship in a bottle with dolphins and lighthouse@vib3coded | — | $12.06 | — | Author’s pick | X post |
| Gemini 4 Argon | Ship in a bottle with dolphins and lighthouse@vib3coded | — | $0.19 | — | — | X post |
| GPT-6.1 Sol | Interactive ship in a bottle@Vectorxeth | 16 min | $10.61 | — | — | X post |
Only numbers stated in each post. “—” means the post doesn’t say.
Summary
2 posts on X run this test with GPT-6.1 Sol, Claude Opus 5.5 and Gemini 4 Argon. @Vectorxeth, @vib3coded report per-model numbers (see the table). According to @Vectorxeth, GPT wins for a quick demo, while Claude built a deeper foundation for a real product. (The author’s view.) According to @vib3coded, Opus for quality, though Gemini did a decent job at almost 64 times cheaper. (The author’s view.)
Who won where
2- Interactive ship in a bottleGPT wins for a quick demo, while Claude built a deeper foundation for a real product.— the author’s view, @Vectorxeth (paraphrased)
- Ship in a bottle with dolphins and lighthouseOpus for quality, though Gemini did a decent job at almost 64 times cheaper.— the author’s view, @vib3coded (paraphrased)
About this test
Posts ask models to build an interactive 3D ship in a glass bottle, with water and waves inside. Results differ most in the glass, the water and how many extras (wind, camera modes, sea life, a lighthouse) the model adds. The short wording above is the prompt @Vectorxeth says he gave both models.
Common wordings
No fixed wording; posts use variations on the same idea.
Build an interactive ship in a bottle.