Reaction
GPT-6.1 Sol's reported token efficiency and near-Astra-level performance
A user says delegating work to Dot reduces token use further and suggests checking Codex for recurring tasks it could handle.
TLDR
A user describes GPT-6.1 Sol as “much more token efficient” with near Astra-level performance so far. They say delegating work to Dot reduces token use further and recommend asking it to find recurring Codex tasks it can handle. They also suggest Sol High as a default and GPT-6 Luna for difficult work.
Combined views
6.4K
1 Source, first seen ago