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-R1 vs Gemma 4 26B A4B

DeepSeekvsGoogle29 shared benchmarks821 head-to-head
BenchmarkDeepSeek-R1Gemma 4 26B A4B
AA Agentic Index3.132.1
AA Intelligence18.626.1
AA-LCR5661.7
AA-Omniscience-31.3-50.8
AIME 202587.590
Artificial Analysis Coding Index24.639.3
browsecomp8.926.3
C-Eval91.882.5
critpt0.60
gdpval1.525.7
GPQA Diamond8182.3
HLE17.719.3
hmmt_feb_202576.791.7
IFBench3972.4
ifeval83.396.4
LiveCodeBench84.479.8
mmlu92.977.8
mmlu_redux93.492.7
MMLU-Pro8585.2
OmniScience Accuracy30.719.1
OmniScience Non-Hallucination10.513.6
scicode35.740.3
SWE-bench Multilingual30.543.4
SWE-bench Verified57.657.4
TauBench V3 - Banking6.412
Terminal-Bench 2.119.139
Terminal-Bench Hard6.125
τ²-Bench Telecom (AA run)11.443.6
τ³-Bench6.459

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.