Same prompt, same skill: Codex beats Gemini in head-to-head video generation test
huangyun_122 · x · 2026-08-18
A blogger gave Gemini and Codex the identical skill and copy to generate a Xiaohongshu account-growth tutorial video, including script and images. Codex won both rounds by a wide margin.
More from Models
- LatentMDM Outperforms AR with KV Caching on TinyGSM — msalbergo · 2026-08-18
- Qwen3.8 27B outscores GPT-5.6-Terra on Artificial Analysis Agentic Index — UnknownEssence · 2026-08-18
- Redditor calls benchmark scores meaningless: 1.6T-param DeepSeek only 1 point ahead of 27B Qwen — Nerfariox · 2026-08-18
- How to host Qwen3.8-27b on a single RTX 5090: quants, context, and trade-offs — DustNearby2848 · 2026-08-18
- Anthropic's watermarking and the shift to AI governance and authority — asusarla · 2026-08-18
- Open 27B models make others 'unreasonable' — Reddit thread argues LLMs are nearing saturation — Potential_Block4598 · 2026-08-18