gemini-4POST

Argon 图像与长视频评测领先 GPT-6 Astra

ScaleAILabs 称 Argon 图像类别 62.8% 首破 60%，领先 GPT-6-astra 的 55.6% 和 Claude Opus 5.5 的 51.3%；10 分钟以上长视频达 59.1%，其他模型不超 41%。

Scale Labs

@ScaleAILabs21 天前收录原推发布于 2026.10.15

Scale Labs @ScaleAILabs原推发布于 2026.10.15

Argon’s biggest edge is in images. At 62.8%, it’s the first model to break 60% in that category, landing roughly 7 points ahead of the next best (GPT-6-astra at 55.6%, Claude Opus 5.5 at 51.3%). The most striking category result comes from long videos. On clips over 10 minutes, Argon reaches 59.1%, while no other model gets past 41%.

x.com/scaleailabs/status/2110455010236780549采集于 2026.10.15 11:35