Head to head · tested 27 Oct 2026

Gemma 4 26B A4B vs Gemma 4 31B: accuracy, speed and price

Gemma 4 31B is more accurate; Gemma 4 26B A4B is cheaper.

GoogleGemma 4 26B A4B15.0% score · #71$0.0011 per question172 tok/s

GoogleGemma 4 31B30.0% score · #64$0.0014 per question58 tok/s

Verdict

Built from the measurements below

Gemma 4 31B is more accurate: 30.0% against 15.0% on the same 80 questions, a 15.0-point gap that is statistically significant.
Gemma 4 26B A4B is 1.3× cheaper per question as billed ($0.0011 vs $0.0014), reasoning tokens included.
Gemma 4 26B A4B generates text 2.9× faster: 172 tokens per second against 58.

Which should you choose?

Choose Gemma 4 26B A4B for faster streaming and more hosts to choose from on OpenRouter (15 against 13). Choose Gemma 4 31B for the most accurate answers of the two and the lower cost per correct answer ($0.0046 against $0.0072). The widest subject gap is in humanities & social science: Gemma 4 31B answered 5 of 9 correctly, Gemma 4 26B A4B 2 of 9. With so few questions per subject, treat this as a lead to test on your own prompts, not a result.

Accuracy gap -15.0 points (z = -2.27). We call a winner only when z is at least 1.96, the usual 95% threshold. Speed, latency and price need a 10% margin.

Head to head

Same questions, same prompt, same API

Measure	Gemma 4 26B A4B	Gemma 4 31B	Edge
Verification scoreHigher is better	15.0%	30.0%	Gemma 4 31B
Cost per questionLower is better	$0.0011	$0.0014	Gemma 4 26B A4B
Output speedHigher is better	172	58	Gemma 4 26B A4B
Time to first tokenLower is better	0.7 s	0.6 s	Similar
End-to-end response timeLower is better	4.8 s	12.6 s	Gemma 4 26B A4B
Tokens per questionLower is better	3,722	3,307	Gemma 4 31B
Context windowHigher is better	262k	262k	Similar

Side by side

Bars are scaled to the larger of the two values. The winner on each measure is marked in bold.

Every measure

Gemma 4 26B A4B first, Gemma 4 31B second

Verification scorehigher is better

Gemma 4 26B A4B15.0%

Gemma 4 31B30.0%

Cost per questionlower is better

Gemma 4 26B A4B$0.0011

Gemma 4 31B$0.0014

Output speedhigher is better

Gemma 4 26B A4B172

Gemma 4 31B58

Time to first tokenlower is better

Gemma 4 26B A4B0.7 s

Gemma 4 31B0.6 s

End-to-end response timelower is better

Gemma 4 26B A4B4.8 s

Gemma 4 31B12.6 s

Tokens per questionlower is better

Gemma 4 26B A4B3,722

Gemma 4 31B3,307

Context windowhigher is better

Gemma 4 26B A4B262k

Gemma 4 31B262k

Score vs. cost per question

Both models in color, every other model in grey

Results by subject and dataset

Questions answered correctly. Small samples: a gap of one or two questions is noise.

Subject	Questions	Gemma 4 26B A4B	Gemma 4 31B
Science	30	4/30	11/30
Engineering	15	0/15	0/15
Medicine	11	1/11	2/11
Economics	9	4/9	5/9
Law	6	1/6	1/6
Humanities & social science	9	2/9	5/9
SuperGPQA dataset	50	1/50	7/50
MMLU-Pro dataset	30	11/30	17/30

Specifications and pricing

From OpenRouter's catalogue, plus our test dates

Specification	Gemma 4 26B A4B	Gemma 4 31B
Creator	Google	Google
Released	3 Apr 2026	2 Apr 2026
Weights	Open weights	Open weights
Reasoning	Yes	Yes
Input	image, text, video	image, text, video
Output	text	text
Context window	262k tokens	262k tokens
Knowledge cutoff	—	—
List price in / out per 1M	$0.09 / $0.3	$0.2 / $0.4
Blended price per 1M (3:1)	$0.14	$0.25
Cheapest host	Darkbloom · $0.042 / $0.22	DeepInfra · $0.16 / $0.32
Hosts on OpenRouter	15	13
Last tested	27 Oct 2026	27 Oct 2026

Compare any two models →

Gemma 4 26B A4B vs Gemma 4 31B: frequently asked questions

Is Gemma 4 26B A4B better than Gemma 4 31B?: Gemma 4 31B is more accurate: it answered 24 of 80 questions correctly (30.0%) against 12 of 80 (15.0%) for Gemma 4 26B A4B. Gemma 4 31B is more accurate; Gemma 4 26B A4B is cheaper.
Which is cheaper, Gemma 4 26B A4B or Gemma 4 31B?: Gemma 4 26B A4B is cheaper: $0.0011 per question as billed, against $0.0014. List prices: Gemma 4 26B A4B $0.09 in / $0.3 out, Gemma 4 31B $0.2 in / $0.4 out per million tokens.
Which is faster, Gemma 4 26B A4B or Gemma 4 31B?: Gemma 4 26B A4B is faster at 172 output tokens per second, against 58. Time to first token: 0.7 s for Gemma 4 26B A4B, 0.6 s for Gemma 4 31B.
Which has the larger context window, Gemma 4 26B A4B or Gemma 4 31B?: Both accept about the same amount of context: 262k and 262k tokens.

Measure

Gemma 4 26B A4B

Gemma 4 31B

Edge

Verification scoreHigher is better

15.0%

30.0%

Gemma 4 31B

Cost per questionLower is better

$0.0011

$0.0014

Gemma 4 26B A4B

Output speedHigher is better

172

Gemma 4 26B A4B

Time to first tokenLower is better

0.7 s

0.6 s

Similar

End-to-end response timeLower is better

4.8 s

12.6 s

Gemma 4 26B A4B

Tokens per questionLower is better

3,722

3,307

Gemma 4 31B

Context windowHigher is better

262k

Similar

Subject

Questions

Gemma 4 26B A4B

Gemma 4 31B

Science

4/30

11/30

Engineering

0/15

Medicine

1/11

2/11

Economics

4/9

5/9

Law

1/6

Humanities & social science

2/9

5/9

SuperGPQA dataset

1/50

7/50

MMLU-Pro dataset

11/30

17/30

Specification

Gemma 4 26B A4B

Gemma 4 31B

Creator

Google

Released

3 Apr 2026

2 Apr 2026

Weights

Open weights

Reasoning

Yes

Input

image, text, video

Output

text

Context window

262k tokens

Knowledge cutoff

—

List price in / out per 1M

$0.09 / $0.3

$0.2 / $0.4

Blended price per 1M (3:1)

$0.14

$0.25

Cheapest host

Darkbloom · $0.042 / $0.22

DeepInfra · $0.16 / $0.32

Hosts on OpenRouter

Last tested

27 Oct 2026

Gemma 4 26B A4B vs Gemma 4 31B: frequently asked questions

Is Gemma 4 26B A4B better than Gemma 4 31B?

Gemma 4 31B is more accurate: it answered 24 of 80 questions correctly (30.0%) against 12 of 80 (15.0%) for Gemma 4 26B A4B. Gemma 4 31B is more accurate; Gemma 4 26B A4B is cheaper.

Which is cheaper, Gemma 4 26B A4B or Gemma 4 31B?

Gemma 4 26B A4B is cheaper: $0.0011 per question as billed, against $0.0014. List prices: Gemma 4 26B A4B $0.09 in / $0.3 out, Gemma 4 31B $0.2 in / $0.4 out per million tokens.

Which is faster, Gemma 4 26B A4B or Gemma 4 31B?

Gemma 4 26B A4B is faster at 172 output tokens per second, against 58. Time to first token: 0.7 s for Gemma 4 26B A4B, 0.6 s for Gemma 4 31B.

Which has the larger context window, Gemma 4 26B A4B or Gemma 4 31B?

Both accept about the same amount of context: 262k and 262k tokens.

Gemma 4 26B A4B vs Gemma 4 31B: accuracy, speed and price

Verdict

Which should you choose?

Head to head

Side by side

Every measure

Score vs. cost per question

Results by subject and dataset

Specifications and pricing

Related comparisons

Gemma 4 26B A4B vs Gemma 4 31B: frequently asked questions

Gemma 4 26B A4B vs Gemma 4 31B: accuracy, speed and price

Verdict

Which should you choose?

Head to head

Side by side

Every measure

Score vs. cost per question

Results by subject and dataset

Specifications and pricing

Related comparisons

Gemma 4 26B A4B vs Gemma 4 31B: frequently asked questions