Gemini 3 Flash
Google · Released 17 Dec 2025 · gemini-3-flash-preview
$0.50
Input price / 1M tokens
#14 of 61, cheapest first
$3.00
Output price / 1M tokens
#19 of 61, cheapest first
1.05M
Context window
#11 of 71
182
Output tokens / second
#14 of 61
33.7%
Humanity’s Last Exam (no tools)
#14 of 23
33.6%
ARC-AGI-2
#6 of 6
90.4%
GPQA Diamond
#16 of 27
95.2%
AIME 2025
#5 of 7
Summary
Gemini 3 Flash is a proprietary model from Google, released on 17 Dec 2025.
At $0.50 input and $3.00 output per million tokens, it is #14 of 61 on input price, cheapest first.
Its 1.05M-token context window ranks #11 of 71.
Measured output speed is 182 tokens per second, #14 of 61.
Legacy Gemini 3 preview Flash model giving baseline speed and intelligence. It is still served, but newer GA Flash models replace it.
Benchmarks · 10 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| Humanity’s Last Exam (no tools)no tools, full set text + multimodal (43.5% with search and code execution) | 33.7% | #14 of 23 | Google Gemini 3 Flash model cardSelf-reported | |
| ARC-AGI-2ARC Prize Verified | 33.6% | #6 of 6 | ARC Prize, cited in Google's model card | |
| GPQA Diamondno tools | 90.4% | #16 of 27 | Google Gemini 3 Flash model cardSelf-reported | |
| AIME 2025no tools (99.7% with code execution) | 95.2% | #5 of 7 | Google Gemini 3 Flash model cardSelf-reported | |
| MMMU-Pro | 81.2% | #2 of 12 | Google Gemini 3 Flash model cardSelf-reported | |
| SWE-bench Verifiedsingle attempt | 78% | #3 of 12 | Google Gemini 3 Flash model cardSelf-reported | |
| Terminal-Bench 2.0Terminus 2 harness | 47.6% | #5 of 7 | Google Gemini 3 Flash model cardSelf-reported | |
| τ²-bench | 90.2% | #3 of 6 | Google Gemini 3 Flash model cardSelf-reported | |
| SimpleQA Verified | 68.7% | – | Google Gemini 3 Flash model cardSelf-reported | |
| MRCR v2 (8-needle)128k average (22.1% at 1M pointwise) | 67.2% | #7 of 13 | Google Gemini 3 Flash model cardSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with Gemini 3 Flash, and the closest scores.
Humanity’s Last Exam (no tools)
#14 of 23Expert questions, no search or code
- Gemini 3 Flash33.7
- Gemini 3.1 Pro44.4
- Gemini 2.5 Pro21.6
- Gemini 2.5 Flash11
- Gemma 4 31B19.5
- Gemma 4 26B A4B8.7
GPQA Diamond
#16 of 27Graduate-level science questions
- Gemini 3 Flash90.4
- Gemini 3.1 Pro94.3
- Gemini 2.5 Pro86.4
- Gemini 2.5 Flash82.8
- Gemma 4 31B84.3
- Gemma 4 26B A4B82.3
Compare Gemini 3 Flash side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- price, cache, batch, free tier: ai.google.dev/gemini-api/docs/pricing, checked 2 Oct 2026
- context, max output, modalities, api id, preview status: ai.google.dev/gemini-api/docs/models/gemini-3-flash-preview, checked 2 Oct 2026
- release date, replacement: ai.google.dev/gemini-api/docs/deprecations, checked 2 Oct 2026
- model card, benchmarks: storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-Flash-Model-Card.pdf, checked 2 Oct 2026
- output speed: artificialanalysis.ai/models/gemini-3-flash, checked 2 Oct 2026
Building on Gemini 3 Flash?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.