Sign up
Back to Model Arena

MODEL ARENA / TEST 01 / EVALUATION

Sky City: how did each result perform?

The four single-file HTML results were checked against the rubric shipped with the original prompt. These are one reviewer's visual observations, browser behavior checks, and raw metrics. Generation conditions and real-device performance are not fully verified, so no overall score or formal ranking is published.

/01

Original prompt

Initial test 01 prompt and its matching rubric

/02

Same browser

Chrome 155 · 1920 × 1080 · DPR 1

/03

Full 60 seconds

Recorded separately without speed changes

Behavior checks

Object counts are visible API candidates at seconds 20 / 25 / 30 / 35, not proof that every item is visually distinguishable. Category counts are the minimum across those samples. All four reached and held the 60-second ending; pause froze time and play resumed it.

ModelObject candidatesCategoriesHover shopsAuto cards
GPT-6 Astra85 / 86 / 86 / 8576/63/3
Claude Opus 5.574 / 75 / 73 / 7076/63/3
Kimi K3.1 Pro70 / 64 / 72 / 7656/63/3
DeepSeek V4.1 Flash110 / 110 / 113 / 11356/63/3

All six shops responded to real pointer hover; three different auto cards appeared for about three seconds each. Touch input and manual priority were also checked.

Observations by model

GPT-6 Astra preview

GPT-6 Astra

Codex · reasoning high

Strengths: The three suspended districts read clearly. Soft daylight and clouds support a coherent panorama, with trains, craft, and a large airship carrying the final event.

Limits: Tower silhouettes repeat, and shops and people are small in the wide shots.

Open original HTML
Claude Opus 5.5 preview

Claude Opus 5.5

Claude Code · reasoning high

Strengths: The six storefronts, props, and pedestrians are especially readable. The ending returns to a city view with a clear title.

Limits: The screen is still heavily covered by white clouds around second 10, and the platform underside and distant buildings are simpler.

Open original HTML
Kimi K3.1 Pro preview

Kimi K3.1 Pro

OpenCode · reasoning high

Strengths: Deck silhouettes, vertical connections, and height above the clouds read well. Shop cards and the airship event arrive on schedule.

Limits: The framing changes little across the sampled moments, and street life and materials remain relatively light.

Open original HTML
DeepSeek V4.1 Flash preview

DeepSeek V4.1 Flash

OpenCode · reasoning max · run 2

Strengths: Facades, platform undersides, bridges, and traffic carry substantial detail. Dusk lighting gives the scene density, and the airship and final title read clearly.

Limits: Some shops and people are small in the wide view; the central scene is busy, and mid-film camera motion is restrained.

Open original HTML

Scoring limits and source files

Headless Chrome reported a 16.7–16.8 ms rAF P95 and 0–0.03% of intervals over 50 ms. That cadence alone cannot establish animation quality on a physical display, so the 10-point performance item remains unscored. A frame-by-frame count of visually distinguishable objects also remains open. The rubric therefore precludes an overall score.

Model labels come from the delivered filenames; only the DeepSeek file marks run 2. Independent records of the exact submitted prompts, attempt counts, and repair opportunities are unavailable, so these results are not treated as a controlled model ranking.