Gemini 3.8 Flash: benchmark results
Google's Gemini 3.8 Flash (September 2, 2026), the fast general-purpose tier of the Gemini 3 family and the successor to Gemini 3.7 Flash, aimed at software engineering and agentic tasks. Provider: Google. Released 2026-09-02. Access: API.
Aggregate Rating 1873 ± 5, rank #21 of 1656 rated models, from 865 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| 3DVerifier - Affordance Region | 77.3 | Balanced accuracy (%) | 100 |
| 3DVerifier - Object | 76.5 | Balanced accuracy (%) | 100 |
| APEX-Harm - Task Utility - Attacker-Provided Generic Script | 0.87 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Dynamic Prompt Injection | 0.88 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Dynamic Script Injection With Permission | 0.88 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Dynamic Script Injection Without Permission | 0.9 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Overall | 0.88 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Static Prompt Injection | 0.87 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Static Script Injection | 0.87 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Task-Specific Replacement Direct Action | 0.88 | Mean Task Score (0–1) | 100 |
| APEX-Harm - Task Utility - Task-Specific Replacement Runtime Information | 0.89 | Mean Task Score (0–1) | 100 |
| BioMysteryBench Human-Difficult | 56.5 | Accuracy (mixed sources) | 100 |
Interactive version: theaggregate.ai/model?slug=gemini-3-8-flash · How It Works · Data refreshed daily, snapshot 2026-10-20.