New AI model GPT-6.2 Sol tops ranking at a fraction of rival's cost

A new AI model called GPT-6.2 Sol has launched in a limited test and landed in first place on a public AI leaderboard, according to BigGo Finance. The bigger surprise is the price. The outlet reports that each task costs just 11.8% as much as it does with Opus 5.5, a rival model from Anthropic.

What changed

GPT-6.2 Sol is rolling out in what is called a gray release. That means a limited test before a full launch, so not everyone can use it yet. A leaderboard is a public ranking where AI models are compared against each other. Sol debuted at the top of one.

Why the cost matters

Per-task cost is what it costs an AI model to finish one job, such as answering a question or drafting a document. If one model does that for about an eighth of the price of another, the savings can add up fast for any company that runs AI all day.

Who it affects

Most people will not pick a model directly. They will feel it through the apps and services they already use. Cheaper models can lower the cost of tools your business already pays for, and they can lead to price cuts or new features.

What is not known yet

The material does not say which leaderboard was used, how the cost was measured, or what tasks were tested. It also gives no date for a full launch, no public pricing and no details on who can try Sol now. The claims come from one outlet, so treat them as early reports.

What happens next

Watch for the move from limited test to full launch, and for any pricing that confirms the savings. Also watch whether rivals such as Anthropic respond with changes of their own.