Gemini 4 Argon (Medium): benchmark results
Provider: Google. Released 2026-09-30. Access: API.
Aggregate Rating 1971 ± 25, rank #6 of 2159 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AutomationBench | 50.08 | Strict Task Pass Rate (%) | 99.1 |
| ARC-AGI-1 | 98.5 | Accuracy (%) | 98.3 |
| Bug Hunt Bench - LMS | 25.7 | Planted Bugs Fixed (out of 60) | 98.3 |
| Argo-Bench - Solved | 34.29 | Share of the 210 tasks solved (%), a task counting as solved | 95.9 |
| Conceptual Reasoning Index - Decision Theory (DTBench) | 95.55 | Chance-Corrected Score (0-100) | 95.7 |
| ARC-AGI-3 | 46.27 | Accuracy (%) | 92.1 |
| CursorBench 4.0 | 52.8 | Score (%) | 91.8 |
| ARC-AGI-2 | 87.08 | Accuracy (%) | 90.8 |
| Conceptual Reasoning Index - Argument Evaluation (LMCA) | 62.85 | Chance-Corrected Score (0-100) | 89.9 |
| Argo-Bench | 51.28 | Mean task score (0-100) over 210 enterprise data-science tas | 89.8 |
| Bug Hunt Bench | 39.3 | Planted Bugs Fixed (out of 105) | 89.6 |
| Conceptual Reasoning Index | 72.06 | Chance-Corrected Score (0-100) | 86.6 |
Interactive version: theaggregate.ai/model?slug=gemini-4-argon-medium · How It Works · Data refreshed daily, snapshot 2026-10-18.