VECTOR WIREAI INTELLIGENCE
NVDA$1,847+3.2%MSFT$512+1.1%GOOGL$199-0.4%META$728+2.7%AMD$184-1.2%TSM$212+0.6%PLTR$98+4.1%AI IDX4,821+1.9%
PKT
SEEDRefresh Models Deals Regulatory Sources

DeepSeek-V4-Pro vs Gemini 2.5 Flash

DeepSeekvsGoogle25 shared benchmarks214 head-to-head
BenchmarkDeepSeek-V4-ProGemini 2.5 Flash
AA Agentic Index63.318.8
AA Intelligence5317.7
AA-LCR7065.7
AA-Omniscience-10.6-29.8
arena_text_factuality14501437
Artificial Analysis Coding Index59.422.2
critpt131.4
gdpval4911.9
Global-MMLU-Lite89.388.4
GPQA Diamond90.582.8
HLE48.212.1
IFBench76.550.3
LiveCodeBench93.576.2
OmniScience Accuracy4326.1
OmniScience Non-Hallucination12.224.7
scicode5035.9
simplebench50.941.2
simpleqa57.926.9
SWE-bench Verified80.660.4
Terminal-Bench Hard46.213.6
vectara_answer_rate97.299
vectara_avg_summary_length153.8101.5
vectara_factual_consistency91.492.2
vectara_hallucination_rate8.67.8
τ²-Bench Telecom (AA run)96.231.6

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.