跳到正文

DeepSeek V4.1 Flash 的 ARC-AGI 成绩

摘要

来自 @deepseek_ai 的 DeepSeek V4.1 Flash 1017 在 ARC-AGI 上的成绩(已核验): - ARC-AGI-3:5.8%,$1.2K(provider adapter harness),0.2%,$1.0K(standard harness) 它在 ARC-AGI-3(5.81%)和 ARC-AGI-2(78.1%)上领先开放权重模型,并在 ARC-AGI-1(94.5%)上并列开放权重模型最佳成绩。

摘要由模型根据原文撰写,以原文为准。

正文 · 原文

DeepSeek V4.1 Flash 1017 from @deepseek_ai on ARC-AGI (Verified):

- ARC-AGI-3: 5.8%, $1.2K (provider adapter harness), 0.2%, $1.0K (standard harness)

It leads open-weight models on ARC-AGI-3 (5.81%) and ARC-AGI-2 (78.1%) and ties the best open-weight score on ARC-AGI-1 (94.5%).

来源:ARC Prize · x.com

引用原文

arcprize. DeepSeek V4.1 Flash 1017 from @deepseek_ai on ARC-AGI (Verified):[EB/OL]. (2026-11-03)[访问日期]. https://x.com/arcprize/status/2117360588959461708.