gemini-3-flash-preview
proprietaryMultimodal
Benchmark Average (0–100)
80.8%
Average of non-null benchmark scores across all evaluated tasks.
BENCHMARK SCORES
GPQA90.4%
Graduate-Level Google-Proof Q&A. PhD-level multiple-choice questions in chemistry, biology, and physics. Scored by accuracy.
AIME 202599.7%
AIME 2025 is a mathematics competition benchmark testing advanced problem-solving.
SWE-bench Verified78.0%
SWE-bench Verified evaluates models on real-world software engineering tasks from GitHub issues.
MMMLU91.8%
Multilingual MMLU tests knowledge across many languages and subject areas.
HLE43.5%
Humanity's Last Exam (HLE) is a multi-modal benchmark testing frontier knowledge across mathematics, humanities, and natural sciences with 2,500 expert-level questions.
MMMU-Pro81.2%
MMMU-Pro is a more challenging version of MMMU with harder multimodal understanding tasks.
OTHER BENCHMARKS
SimpleQA
68.7%
ToolAthalon
49.4%
ARC-AGI v2
33.6%
CharXIV-R
80.3%
ScreenSpot Pro
69.1%
MCP Atlas
57.4%
MODEL INFO
Organization
Google
Context Window
1,000,000
Release Date
2025-12-17
Knowledge Cutoff
2025-01-31
Parameters
—
License
proprietary
Input Price ($/M)
$0.5000
Output Price ($/M)
$3.0000
METADATA
Announcement Date
2025-12-17
Organization Country
US
Canonical ID
—