Google returns to first place in independent artificial intelligence model ranking

Artificial Analysis, an organization that publishes independent rankings on the performance of artificial intelligence models, has released the fifth version of its Intelligence Index, revising the criteria for comparing systems from the leading laboratories in the sector.

In the new ranking, Google takes first place with 59 points, followed by Claude Opus 5.5, developed by Anthropic, with 58 points, and GPT-6 Astra by OpenAI with 57 points. The margin between the top three positions is therefore minimal, confirming a phase in which major laboratories are producing systems with very close performance on the standardized tests used by the index.

The news currently comes from a single primary source — the Artificial Analysis report — picked up in substantially identical form by two outlets that reported on it; since there is no independent confirmation of the data beyond the source that produced it, the result should be read as the outcome of a single measurement methodology, not as a cross-check among multiple parties.

This limitation is not secondary. Indices of this kind are based on batteries of tests chosen and weighted by those who build them: a change in methodology, such as the one noted with the move to version 5, can by itself shift the rankings without the underlying models having changed. There is currently no second body measuring the same performance with its own independent criteria, so Google’s lead should be considered relative to this specific evaluation tool.

One verifiable fact remains: three systems from three different companies — Google, Anthropic and OpenAI — are contending for the top spot with a minimal score gap, a sign that none of the three laboratories currently holds a stable advantage according to this index.