Gemini 3
Google·
LLMsflagshipcloud
Overview
Next major generation with agentic-first architecture.
Capabilities and innovations
Agentic-firstInfinite contextPersonalizationAntigravity IDENative Agentic Workflow
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 91.9% | unsourced |
| Humanity’s Last Exam (no tools) | 41.2% | unsourced |
| SWE-bench Verified | 72.5% | unsourced |
| MMLU | 93.1% | unsourced |
| MMLU-Pro | 90.1% | unsourced |
- SWE-bench Pro: not reported by the vendor
- Terminal-Bench 2.x: benchmark did not exist at release
- MMMU-Pro: benchmark did not exist at release
Reliability
- Hallucination rate (Vectara HHEM)
- 13.6% (lower is better)independent
- Agentic tool use (τ-bench)
- 75.3% (higher is better)third party
Links
More from Google
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.