Four models animate a pelican driving a flying Roadster
MiniMax M3.1 Flash, DeepSeek V4.1 Flash, Claude Haiku 5.5 and GPT-6 Luna each generated the same pelican-in-a-flying-Roadster animation, compared on time, tokens and cost, with Haiku 5.5's pelican faring worst.
MINIMAX M3.1 FLASH OUTBUILT DEEPSEEK, CLAUDE AND GPT? 😈 One task: a pelican driving a flying Roadster MiniMax M3.1 Flash → 43m 54s · 41M tokens · $0.62 DeepSeek V4.1 Flash → 1h 09m 22s · 54.68M tokens · $0.51 Haiku 5.5 → 30m 23s · 5.81M tokens · $0.65 GPT‑6 Luna → 17m 40s · 7.92M tokens · $0.14 MiniMax nailed the air jets. DeepSeek had the cleanest animations. GPT finished first and cheapest. Claude Haiku’s pelican… needs some work 😭 Is MiniMax already pushing its way into the top tier of Flash models?
This post quotes another post: “Gemini 4 Argon vs GPT-6 Astra vs Opus 5.5: same task time and cost”