VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Refresh Models —Deals —Regulatory —Sources —

Gemini 3.7 Flash vs Gemini 3.8 Flash

30 SHARED BENCHMARKS

Across 30 shared benchmarks, Gemini 3.7 Flash scores higher on 8 and Gemini 3.8 Flash on 22. The widest gap is OmniScience Non-Hallucination, where Gemini 3.8 Flash scores 48.1 against 35.5.

GOOGLEVSGOOGLE30 SHARED8–22 HEAD-TO-HEAD
BenchmarkGemini 3.7 FlashMARGINGemini 3.8 Flash
AA Agentic Index45.150
AA Intelligence5659
AA-LCR8182
AA-Omniscience26.529.6
arena_elo14911494
Artificial Analysis Coding Index76.176.3
charxiv_rq88.786.2
critpt14.318.3
DeepSWE 1.165.373.7
GDP (Surge AI)3435
gdpval50.852.3
GDPval-AA v215251545
GPQA Diamond94.595.3
HLE47.947.8
HLE-Verified53.654.9
livebench_agentic_coding58.354.2
livebench_coding78.972.5
livebench_data_analysis6854
livebench_instruction_following79.981.4
livebench_language85.587.8
livebench_math93.591.6
livebench_reasoning87.889.3
LVBench85.487.1
MMMU-Pro85.585.6
OmniScience Accuracy55.354.6
OmniScience Non-Hallucination35.548.1
OSWorld 2.047.959
scicode57.954.4
TauBench V3 - Banking35.545.8
Terminal-Bench 2.185.889.4

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.