ODE Identifiability Complete Class Recall Domain-Anonymized - Nonidentifiable: leaderboard
ODE Identifiability Complete Class Recall Domain-Anonymized - Nonidentifiable: Parameter classification Nonidentifiable (%). ODE identifiability arXiv 2611.14184v1: 88 scenarios, 696 parameter-level references, ten repeated stateless requests/condition. Exact ASU gateway IDs and completion denominators are pinned in S2 Table. API observations retain gateway identity instead of asserting a provider-pinned backend. Complete six-model comparison S4 stays separate from incomplete exploratory Table 2; Domain-Aware and Domain-Anonymized prompts separate. Accuracy means ten-run percentages. FPR/FNR are false-positive/false-negative rates, lower is better; class recall is higher. GPT thinking medium/high explicit reasoning_effort; other provider defaults unreported.
Metric: Parameter classification Nonidentifiable (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 98.35 | #91 |
| 2 | Claude Opus 4.5 | 88.61 | #152 |
| 3 | Claude Sonnet 4.5 | 83.76 | #244 |
| 4 | Gemini 3.5 Flash | 82.99 | #98 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ode-identifiability-complete-class-recall-domain-anonymized-nonidentifiable · How It Works · Data refreshed daily, snapshot 2026-11-12.