Is Gemma 4 26B A4B better than Granite 4.0 Micro?

On accuracy they are statistically tied: Gemma 4 26B A4B scored 15.0% and Granite 4.0 Micro 25.0% on the same 80 questions from open datasets. Similar accuracy; Granite 4.0 Micro costs less per question.

Which is cheaper, Gemma 4 26B A4B or Granite 4.0 Micro?

Granite 4.0 Micro is cheaper: $0.000051 per question as billed, against $0.0011. List prices: Gemma 4 26B A4B $0.09 in / $0.3 out, Granite 4.0 Micro $0.017 in / $0.112 out per million tokens.

Which is faster, Gemma 4 26B A4B or Granite 4.0 Micro?

Gemma 4 26B A4B is faster at 172 output tokens per second, against 38. Time to first token: 0.7 s for Gemma 4 26B A4B, 0.7 s for Granite 4.0 Micro.

Which has the larger context window, Gemma 4 26B A4B or Granite 4.0 Micro?

Gemma 4 26B A4B, with 262k tokens against 131k.

Head to head · tested 27 Oct 2026

Gemma 4 26B A4B vs Granite 4.0 Micro: accuracy, speed and price

Similar accuracy; Granite 4.0 Micro costs less per question.

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

IBMGranite 4.0 Micro25.0% score · #69$0.000051 per question38 tok/s

Verdict

Built from the measurements below

Accuracy is too close to call: Gemma 4 26B A4B scored 15.0% and Granite 4.0 Micro 25.0% on the same 80 questions, a gap within what chance alone could produce.
Granite 4.0 Micro is 21× cheaper per question as billed ($0.000051 vs $0.0011), reasoning tokens included.
Gemma 4 26B A4B generates text 4.5× faster: 172 tokens per second against 38.
Gemma 4 26B A4B accepts more context: 262k tokens against 131k.

Which should you choose?

Choose Gemma 4 26B A4B for faster streaming, a 262k-token context window and more hosts to choose from on OpenRouter (15 against 1). Choose Granite 4.0 Micro for the lower cost per correct answer ($0.0002 against $0.0072) and shorter outputs (about 403 tokens per question against 3,722). The widest subject gap is in humanities & social science: Granite 4.0 Micro 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 -10.0 points (z = -1.58). A winner is called 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	Granite 4.0 Micro	Edge
Verification scoreHigher is better	15.0%	25.0%	Too close to call
Cost per questionLower is better	$0.0011	$0.000051	Granite 4.0 Micro
Output speedHigher is better	172	38	Gemma 4 26B A4B
Time to first tokenLower is better	0.7 s	0.7 s	Similar
End-to-end response timeLower is better	4.8 s	8.9 s	Gemma 4 26B A4B
Tokens per questionLower is better	3,722	403	Granite 4.0 Micro
Context windowHigher is better	262k	131k	Gemma 4 26B A4B

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, Granite 4.0 Micro second

Verification scorehigher is better

Gemma 4 26B A4B15.0%

Granite 4.0 Micro25.0%

Cost per questionlower is better

Gemma 4 26B A4B$0.0011

Granite 4.0 Micro$0.000051

Output speedhigher is better

Gemma 4 26B A4B172

Granite 4.0 Micro38

Time to first tokenlower is better

Gemma 4 26B A4B0.7 s

Granite 4.0 Micro0.7 s

End-to-end response timelower is better

Gemma 4 26B A4B4.8 s

Granite 4.0 Micro8.9 s

Tokens per questionlower is better

Gemma 4 26B A4B3,722

Granite 4.0 Micro403

Context windowhigher is better

Gemma 4 26B A4B262k

Granite 4.0 Micro131k

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	Granite 4.0 Micro
Science	30	4/30	7/30
Engineering	15	0/15	1/15
Medicine	11	1/11	4/11
Economics	9	4/9	2/9
Law	6	1/6	1/6
Humanities & social science	9	2/9	5/9
SuperGPQA dataset	50	1/50	8/50
MMLU-Pro dataset	30	11/30	12/30

Specifications and pricing

From OpenRouter's catalogue, plus the leaderboard's test dates

Specification	Gemma 4 26B A4B	Granite 4.0 Micro
Creator	Google	IBM
Released	3 Apr 2026	20 Oct 2025
Weights	Open weights	Open weights
Reasoning	Yes	No
Input	image, text, video	text
Output	text	text
Context window	262k tokens	131k tokens
Knowledge cutoff	—	—
List price in / out per 1M	$0.09 / $0.3	$0.017 / $0.112
Blended price per 1M (3:1)	$0.14	$0.04
Cheapest host	Darkbloom · $0.042 / $0.22	Cloudflare · $0.017 / $0.112
Hosts on OpenRouter	15	1
Last tested	27 Oct 2026	26 Oct 2026

Compare any two models →

Gemma 4 26B A4B vs Granite 4.0 Micro: frequently asked questions

Is Gemma 4 26B A4B better than Granite 4.0 Micro?: On accuracy they are statistically tied: Gemma 4 26B A4B scored 15.0% and Granite 4.0 Micro 25.0% on the same 80 questions from open datasets. Similar accuracy; Granite 4.0 Micro costs less per question.
Which is cheaper, Gemma 4 26B A4B or Granite 4.0 Micro?: Granite 4.0 Micro is cheaper: $0.000051 per question as billed, against $0.0011. List prices: Gemma 4 26B A4B $0.09 in / $0.3 out, Granite 4.0 Micro $0.017 in / $0.112 out per million tokens.
Which is faster, Gemma 4 26B A4B or Granite 4.0 Micro?: Gemma 4 26B A4B is faster at 172 output tokens per second, against 38. Time to first token: 0.7 s for Gemma 4 26B A4B, 0.7 s for Granite 4.0 Micro.
Which has the larger context window, Gemma 4 26B A4B or Granite 4.0 Micro?: Gemma 4 26B A4B, with 262k tokens against 131k.