DeepSeek V4.1 Flash versus GPT 5.6 Terra and Sol in coding and Catan
A user endorses a quoted comparison placing DeepSeek V4.1 Flash between Terra and Sol in subjective agentic coding performance and Catan, while costing much less than Terra.
TLDR
A post agrees with a quoted assessment that DeepSeek V4.1 Flash sits between GPT 5.6 Terra and Sol in subjective agentic coding performance and Catan, while costing much less than Terra. The author also wants to see ARC-AGI 2 results and criticizes evaluators for being “strangely indifferent” to the model.
Combined views
8.9K
1 Source, first seen ago