deepseek-v4.1VIDEO
DeepSeek V4.1 Flash vs GPT-6 Astra: near-equal output at about 1/10 the cost
At max reasoning, Astra took 20 minutes and $7 while DeepSeek took 1 hour 40 minutes and $0.9; the author concedes Astra is more refined but says the Flash model nearly matches it at about one-tenth the cost.
Leo · Building w/ AI
@leo114119Archived 11 days agoOriginally posted 2026.10.17
Leo · Building w/ AI @leo114119Originally posted 2026.10.17
又一个 Gpt 6 astra 和 deepseek v4.1-1017 flash的对比作品,在最高推理设置下 > astra 用了 20 分钟,成本 $7 > deepseek 用了 1 小时 40 分钟,成本 $0.9 我承认 Gpt 6 Astra 做的更精细。。。但是 一个 Flash模型 与一个旗舰模型 只用 1/10 的价格,几乎完成了同等水平的作品。。 这还需要证明什么吗。。 https://x.com/notjazii/status/2111446763475435807/video/1
x.com/leo114119/status/2111467514312204561Archived 2026.10.18 16:36