OpenAI's GPT-6.1 Terra halves Sol's token price. Compare models on what a finished task costs.

OpenAI shipped GPT-6.1 Terra on October 26 at half the token price of GPT-6.1 Sol, and the number that matters for agents is what a finished task costs. OpenAI's launch post lists $1 per million input tokens and $5 per million output tokens.

The better claim sits further down. At max effort, OpenAI says Terra is over three times as token efficient as Anthropic's Claude Opus 5.5 and one-fourth its price per token, resulting in a ~9x lower cost per task.

That's the gap a price table hides. Two models can bill alike per token and land far apart per job, because one finishes in fewer tokens. These are OpenAI's own benchmark figures, so they're a reason to test, not a verdict.

Here's what I'd do. Take a batch of jobs your agent already runs, the dull ones that fill a day. Run them on both models. Divide the bill by the jobs that came back right, not the jobs attempted. A cheap model that fails twice isn't cheap. OpenAI says as much: for the most complex coding work, Sol and GPT-6 Astra stay its pick.

Per token is the sticker. Per task is the receipt.