Gemini 3 Flash (Preview) (Thinking): benchmark results
Provider: Google. Released 2025-12-17. Access: API.
Aggregate Rating 1707 ± 13, rank #330 of 2145 rated models, from 292 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AgentProcessBench | 81.6 | StepAcc (%): share of assistant steps, pooled over trajector | 100 |
| AgentProcessBench - BFCL First-Error Accuracy | 64 | FirstErrAcc (%): share of trajectories where the judge locat | 100 |
| AgentProcessBench - BFCL Step Accuracy | 81.8 | StepAcc (%): share of assistant steps, pooled over trajector | 100 |
| AgentProcessBench - First-Error Accuracy | 65.8 | FirstErrAcc (%): share of trajectories where the judge locat | 100 |
| AgentProcessBench - GAIA First-Error Accuracy | 65.2 | FirstErrAcc (%): share of trajectories where the judge locat | 100 |
| AgentProcessBench - GAIA Step Accuracy | 79.7 | StepAcc (%): share of assistant steps, pooled over trajector | 100 |
| AgentProcessBench - tau2-Bench Step Accuracy | 83.4 | StepAcc (%): share of assistant steps, pooled over trajector | 100 |
| EuroEval Albanian Knowledge | 96.46 | Knowledge Average Score (%) | 100 |
| EuroEval Catalan Common Sense Reasoning | 80.44 | Common Sense Reasoning Average Score (%) | 100 |
| EuroEval Catalan Knowledge | 87.01 | Knowledge Average Score (%) | 100 |
| EuroEval Croatian Knowledge | 90.14 | Knowledge Average Score (%) | 100 |
| EuroEval Czech NLU | 63.55 | NLU Average Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gemini-3-flash-preview-thinking · How It Works · Data refreshed daily, snapshot 2026-10-16.