Gemini 4 Carbon (Medium): benchmark results
Provider: Google. Released 2026-10-28. Access: API.
Aggregate Rating 1985 ± 14, rank #9 of 2201 rated models, from 75 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA GDP.pdf - Insurance - Criterion Pass Rate | 90.93 | Criterion Pass Rate (%) | 100 |
| AA MMMU-Pro | 89.25 | Accuracy (%) | 100 |
| AA-Omniscience v1.1 Hallucination Rate | 4.24 | Hallucination Rate (%) | 100 |
| AutomationBench-AA | 77.6 | Guardrail-Adjusted Objective Completion (%) | 99.6 |
| AutomationBench-AA - Finance - Raw Objective Completion | 95.95 | Raw Objective Completion (%) | 99.6 |
| AutomationBench-AA - Operations - Raw Objective Completion | 97.39 | Raw Objective Completion (%) | 99.6 |
| AutomationBench-AA - Sales - Raw Objective Completion | 93.65 | Raw Objective Completion (%) | 99.6 |
| AutomationBench-AA - Strict | 54.19 | Strict Task Success (%) | 99.6 |
| AA-Briefcase - PPTX - Rubric Pass Rate | 64.08 | Rubric Pass Rate (%) | 99.4 |
| ARC-AGI-2 | 94.17 | Accuracy (%) | 99.2 |
| AutomationBench-AA - Marketing - Raw Objective Completion | 94.25 | Raw Objective Completion (%) | 99.2 |
| AutomationBench-AA - Support - Raw Objective Completion | 94.03 | Raw Objective Completion (%) | 99.2 |
Interactive version: theaggregate.ai/model?slug=gemini-4-carbon-medium · How It Works · Data refreshed daily, snapshot 2026-11-12.