If significant evidence is not found within 2 months I will conclude this market as “Neither”.

I will not trade in this market.

Resolution criteria

This market resolves to the Gemini model that is officially confirmed or credibly reported to feature recurrent depth architecture (or dynamic recurrent compute layers) analogous to the recurrent depth designs seen in frontier competitors like GPT-6 Astra.

• Carbon: Resolves if official Google/DeepMind documentation, model cards, research papers, or credible technical reporting verifies that Gemini 4 Carbon incorporates recurrent depth architecture.

• Argon: Resolves if official Google/DeepMind documentation, model cards, research papers, or credible technical reporting verifies that Gemini 4 Argon incorporates recurrent depth architecture.

• Neither: Resolves if credible confirmation or reporting indicates that neither model employs recurrent depth, or if significant technical evidence/official confirmation identifying recurrent depth in either model is not found within 2 months of market creation.

Source of Truth:

• Official Google DeepMind announcements, technical reports, and model cards (e.g., DeepMind Research).

• Credible technical journalism and analysis (e.g., SemiAnalysis, The Verge, Ars Technica, or peer-reviewed papers).

Background

Google recently expanded its frontier lineup with Gemini 4 Argon and Gemini 4 Carbon. Speculation in AI architecture discussions has centered on whether recent frontier releases use recurrent depth—a mechanism allowing transformers to dynamically reuse layers or loop computations across recurrent depth steps—similar to architectural discussions surrounding OpenAI's GPT-6 Astra. Because Google often keeps exact architectural details confidential during early rollouts, resolution relies on verified technical disclosures or credible analysis emerging within the specified 2-month window.