Paper 2602.11144
GENIUS: Generative Fluid Intelligence Evaluation Suite
- Published
- Feb 2026
- Research lab
- Independent
- Citations
- 12
- GitHub
- 43 stars
01 In brief
Summary
The paper introduces GENIUS, the first benchmark for evaluating Generative Fluid Intelligence (GFI) in Unified Multimodal Models (UMMs), distinguishing it from Crystallized Intelligence (CI).
GFI is formalized into three primitives: Inducing Implicit Patterns, Executing Ad-hoc Constraints, and Adapting to Contextual Knowledge.
The benchmark comprises 510 expert-curated samples across five tasks and 20 sub-tasks, evaluated with a hybrid protocol using Rule Compliance, Visual Consistency, and Aesthetic Quality metrics, with Gemini-3-Pro as the judge.
Evaluation of 12 models shows significant deficits, with the best model (Nano Banana Pro) scoring only 57.19, and open-source models like Bagel scoring 26.74.
Diagnostic analysis reveals an 'execution gap' where models understand context but fail to generate compliant outputs, and attention visualization shows irregular distributions.
The authors propose a training-free attention intervention strategy that improves Bagel's overall score by 6.18%, validating the theoretical framework that links attention distribution to implicit gradient descent.
The benchmark aims to guide UMMs beyond knowledge utilization toward dynamic, general-purpose reasoning.
02 From the paper
Abstract
Unified Multimodal Models (UMMs) have shown remarkable progress in visual generation. Yet, existing benchmarks predominantly assess $\textit{Crystallized Intelligence}$, which relies on recalling accumulated knowledge and learned schemas. This focus overlooks $\textit{Generative Fluid Intelligence (GFI)}$: the capacity to induce patterns, reason through constraints, and adapt to novel scenarios on the fly. To rigorously assess this capability, we introduce $\textbf{GENIUS}$ ($\textbf{GEN}$ Fluid $\textbf{I}$ntelligence Eval$\textbf{U}$ation $\textbf{S}$uite). We formalize $\textit{GFI}$ as a synthesis of three primitives. These include $\textit{Inducing Implicit Patterns}$ (e.g., inferring personalized visual preferences), $\textit{Executing Ad-hoc Constraints}$ (e.g., visualizing abstract metaphors), and $\textit{Adapting to Contextual Knowledge}$ (e.g., simulating counter-intuitive physics). Collectively, these primitives challenge models to solve problems grounded entirely in the immediate context. Our systematic evaluation of 12 representative models reveals significant performance deficits in these tasks. Crucially, our diagnostic analysis disentangles these failure modes. It demonstrates that deficits stem from limited context comprehension rather than insufficient intrinsic generative capability. To bridge this gap, we propose a training-free attention intervention strategy. Ultimately, $\textbf{GENIUS}$ establishes a rigorous standard for $\textit{GFI}$, guiding the field beyond knowledge utilization toward dynamic, general-purpose reasoning. Our dataset and code will be released at: $\href{https://github.com/arctanxarc/GENIUS}{https://github.com/arctanxarc/GENIUS}$.