DataLearner logo

Gemini 3.7 Flash Benchmark Details

Gemini 3.7 Flash currently shows benchmark results led by Terminal-Bench 2.1 (8 / 41, score 85.80), GDM-MRCR v2 (8-needle, 128K) (1 / 4, score 97), CharXiv RQ (4 / 15, score 88.70).

Benchmark Results

Gemini 3.7 Flash

Benchmark Results

Thinking
Tool usage
Internet

AI Agent - Tool Usage

7 evaluations
Benchmark / mode
Score
Rank/total
BioMysteryBench (Human-solvable)
MediumToolsInternet
87.10
1 / 1
Terminal-Bench 2.1
Thinking EnabledTools
85.80
8 / 41
LABBench2
MediumToolsInternet
82.10
1 / 1
OSWorld 2.0
MediumTools
47.90
4 / 4
BioMysteryBench (Human-difficult)
MediumToolsInternet
43.50
1 / 1
Automation Bench
Thinking EnabledTools
30.40
3 / 6
Terminal-Bench 3.0
HighTools
14.90
3 / 3

Coding and Software Engineer

3 evaluations
Benchmark / mode
Score
Rank/total
Code Arena WebDev
Thinking EnabledTools
1588
1 / 1
DeepSWE
HighTools
65.30
10 / 25
FrontierCode 1.1
Thinking EnabledTools
43.60
2 / 2

General Knowledge

2 evaluations
Benchmark / mode
Score
Rank/total
AA Intelligence Index
Thinking Enabled
56
3 / 5
HLE-Verified
Medium
53.60
1 / 1

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
Agents' Last Exam
MediumTools
26.30
7 / 9

Productivity Knowledge

2 evaluations
Benchmark / mode
Score
Rank/total
GDPval-AA v2
Thinking Enabled
1525
5 / 9
Harvey Lab-AA
Thinking Enabled
90.70
2 / 3

Multimodal Understanding

4 evaluations
Benchmark / mode
Score
Rank/total
CharXiv RQ
Medium
84.50
10 / 15
CharXiv RQ
MediumTools
88.70
4 / 15
LVBench
Medium
85.40
1 / 1
GDP.pdf
Medium
34
1 / 1

Other

1 evaluations
Benchmark / mode
Score
Rank/total
Compare with other models