The Growing Pains of Frontier Models: When Leaderboards Stop Separating and What to Measure Next

TL;DR

This paper analyzes the interactions and trade-offs among frontier AI models' capabilities, proposing diagnostic tools and a playbook to guide measurement and development strategies for the next phase of AI progress.

Contribution

It introduces a decomposition method to diagnose capability emphasis, reveals how cooperation varies across labs and over time, and offers a practical playbook and dashboard for guiding future frontier model evaluations.

Findings

Capabilities tend to cooperate with a correlation of +0.72.

A second capability transition occurs at 30--72B parameters.

SWE-bench saturation indicates the need for new axes of measurement.

Abstract

Leaderboards rank frontier models on independent axes but do not reveal whether capabilities reinforce or trade off across releases -- and at the frontier, this interaction is the more informative signal. We decompose paired SWE-bench and GPQA Diamond scores into a population coupling trend and per-release residual ($h$-field) that diagnoses capability emphasis and identifies which measurement or stress test is most informative next. Across 34 models from 10 labs (2024--2026), capabilities cooperate ($r = +0.72$, $p < 10^{-6}$), but cooperation varies by lab and over time: DeepSeek reversed from reasoning-rich to coding-first ($h$: $+11.2 \to -4.7$, 15.9-pp swing); Google maintains consistent reasoning emphasis; Anthropic oscillates between coding excursions and recovery. Cooperation is not static -- it cascades. Six open-weight architectures confirm a second capability transition at 30--72B, and SWE-bench is now saturating while HLE and instruction-following retain discriminatory spread -- signaling the next axis rotation. We provide a three-level playbook (locate, diagnose, rotate), a per-lab measurement-priority table, and seven falsifiable predictions with timestamped criteria for the next 12 months of frontier releases. Per-lab coupling slopes vary $5\times$ (Google $1.15$ vs. DeepSeek $0.23$), quantifying how efficiently each recipe converts coding gains into reasoning. Five April 2026 releases confirm the diagnostic out of sample ($r$ rises from $+0.72$ to $+0.75$). An interactive dashboard provides phase classification with actionable recommendations, $h$-field diagnostics, per-lab coupling trajectories, ODE-based scaling predictions, benchmark rotation guidance, self-steering demo, and live tracking of all seven predictions: https://zehenlabs.com/cape/.