Google

Gemma 3 12B

Multimodal
Zero-eval
#1VQAv2 (val)
#2MMMU (val)
#2WMT24++
+2 more

by Google

+
+
+
+
About

Gemma 3 12B was developed as part of the third generation of Google's open-source model family, designed to provide enhanced capabilities in a mid-sized format. Built with improved architecture and training techniques, it balances performance with practical deployment considerations for diverse use cases.

+
+
+
+
Pricing Range
Input (per 1M)$0.05 -$0.05
Output (per 1M)$0.10 -$0.10
Providers1
+
+
+
+
Timeline
AnnouncedMar 12, 2025
ReleasedMar 12, 2025
+
+
+
+
Specifications
Training Tokens12.0T
Capabilities
Multimodal
+
+
+
+
License & Family
License
Gemma
Performance Overview
Performance metrics and category breakdown

Overall Performance

26 benchmarks
Average Score
62.5%
Best Score
94.4%
High Performers (80%+)
8

Performance Metrics

Max Context Window
262.1K
Avg Throughput
33.0 tok/s
Avg Latency
0ms
+
+
+
+
All Benchmark Results for Gemma 3 12B
Complete list of benchmark scores with detailed information
GSM8k
text
0.94
94.4%
Self-reported
IFEval
text
0.89
88.9%
Self-reported
DocVQA
multimodal
0.87
87.1%
Self-reported
BIG-Bench Hard
text
0.86
85.7%
Self-reported
HumanEval
text
0.85
85.4%
Self-reported
AI2D
multimodal
0.84
84.2%
Self-reported
MATH
text
0.84
83.8%
Self-reported
Natural2Code
text
0.81
80.7%
Self-reported
FACTS Grounding
text
0.76
75.8%
Self-reported
ChartQA
multimodal
0.76
75.7%
Self-reported
Showing 1 to 10 of 26 benchmarks
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+