Gemini 3 Flash Preview (Reasoning) is a Google Gemini 3 family model, from the generation Google introduced for stronger reasoning, multimodal understanding, coding, and agentic product experiences. This profile focuses on the exact preview or reasoning variant captured in the local benchmark snapshot.
Introducing Gemini 3Queryable facts extracted from the upstream model payload.
Rank #2 across 343 models.
Rank #3 across 269 models.
Rank #4 across 345 models.
Rank #244 across 293 models.
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Higher bars mean stronger relative placement.
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 46.4 | #38 |
| Artificial Analysis Coding Index | coding | 42.6 | #37 |
| Artificial Analysis Math Index | math | 97.0 | #3 |
| MMLU-Pro | reasoning | 89.0% | #4 |
| reasoning |
| 89.8% |
| #15 |
| Humanity's Last Exam | reasoning | 34.7% | #14 |
| LiveCodeBench | coding | 90.8% | #2 |
| SciCode | coding, reasoning | 50.6% | #15 |
| Output Speed | speed | 193.2 tok/s | #28 |
| Time to First Token | speed | 5.61s | #244 |
| Blended Price | cost | $1.13/M | #184 |
| Input Price | cost | $0.500/M | #164 |
| Output Price | cost | $3.00/M | #194 |
| Value Index | cost, overall | 41.2 | #112 |