Anthropic’s Strongest Model is Kept In House
In the August 2026 internal risk report, Anthropic mentions that they have access to a model that’s better than their public models: Model 2. This model is already used to write most of Anthropic’s production code, even though it’s never been released externally.
The Anthropic August 2026 internal risk report mentions that they have access to a model called Model 2 that is better than any of their public models. They say it is already deployed inside their production systems and is used to write the majority of their code. It’s also about 1.5 points better than Claude Mythos 5 on Anthropic’s internal capability index (AECI).
They performed some safety checks before deploying this model, but it was tested less than Mythos 5 outside of the company. They did not find any new instances of misalignment and consider the risk to be low. They do not plan to release this model externally. It’s interesting because the most powerful models in the world’s best AI safety research group aren’t the same models released to the public.
Source: the-decoder.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles