Verdict
Built from the measurements below
- Gemini 4 Argon is more accurate: 84.8% against 30.0% on the same 79 questions, a 54.8-point gap that is statistically significant.
- Gemma 4 31B is 4.5× cheaper per question as billed ($0.0014 vs $0.0063), reasoning tokens included.
- Gemini 4 Argon generates text 1.2× faster: 69 tokens per second against 58.
- Gemma 4 31B starts answering sooner: 0.6 s to the first token, against 3.3 s for Gemini 4 Argon.
- Gemini 4 Argon accepts more context: 1M tokens against 262k.
Which should you choose?
Choose Gemini 4 Argon for the most accurate answers of the two, faster streaming, shorter outputs (about 558 tokens per question against 3,307) and a 1M-token context window. Choose Gemma 4 31B for the lower cost per correct answer ($0.0046 against $0.0074), a quicker first token, open weights you can run yourself and more hosts to choose from on OpenRouter (13 against 2). The widest subject gap is in engineering: Gemini 4 Argon answered 13 of 14 correctly, Gemma 4 31B 0 of 15. With so few questions per subject, treat this as a lead to test on your own prompts, not a result.
Accuracy gap +54.8 points (z = 6.98). We call a winner only when z is at least 1.96, the usual 95% threshold. Speed, latency and price need a 10% margin.