GPT-6.1 Terra

Commenters questioned where GPT-6.1 Terra fits among Luna, Terra, Sol and Astra, especially because newer, cheaper models can outperform older, larger ones on particular tasks. Some found the naming and shifting lineup confusing; others said the tiers still indicate relative size and price within a generation, while capability varies by task. Several doubted Terra’s value, arguing that Luna is cheaper and Sol offers better performance for the cost, or that benchmark results do not show a compelling reason to choose Terra. Others saw a useful middle option for work that needs more reliability than Luna but does not justify Sol.

An OpenAI employee described Terra as a faster, cheaper midrange model, particularly strong in technical communication and competitive on vision and computer use, while generally trailing newer or larger models elsewhere. They argued that model size, reasoning effort and other capabilities are distinct, so a simple ranking cannot capture every tradeoff. Some commenters welcomed OpenAI’s focus on cost per task and thought Terra could suit routine, well-specified engineering work; others criticized its benchmark showing and called for a stronger Astra release to compete with top-tier rivals. There was also skepticism toward price-performance charts, with one commenter arguing that results depend heavily on which tasks and costs are measured.