Which is better, Gemma 4 26B A4B or NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16?

Refresh Models —Deals —Regulatory —Sources —

Questions this page answers

Which is better, Gemma 4 26B A4B or NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16?: Across 16 shared benchmarks, Gemma 4 26B A4B scores higher on 5 and NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 on 11. The widest gap is Minerva Math, where NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 scores 82.8 against 43.7.

Gemma 4 26B A4B vs NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16

Across 16 shared benchmarks, Gemma 4 26B A4B scores higher on 5 and NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 on 11. The widest gap is Minerva Math, where NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 scores 82.8 against 43.7.

GooglevsNVIDIA16 shared benchmarks5–11 head-to-head

Benchmark	Gemma 4 26B A4B	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16
AGIEval-En	55.3	70
arc_challenge	92.8	92.7
Global-MMLU-Lite (5-shot) Average	74.8	75.5
GSM8K	77	91.3
hellaswag	85.3	85.5
humaneval	50	77.4
MBPP	68.4	78.6
Minerva Math	43.7	82.8
mmlu	77.8	78.6
MMLU-Pro	85.2	67.9
OpenBookQA	48.8	47.6
PIQA (acc-norm)	83.8	85.2
PIQA (acc)	82.2	83.3
RULER 1M	72.9	69.6
RULER 256K	85.7	76.9
winogrande	79.1	80

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.