Models / Gemma 4 26B A4B

Google

Gemma 4 26B A4B

33 published results from 4 sources. Each card shows where the number comes from and what it does not measure. The overall leaderboard combines them; here each stands alone.

Provider
Google
Sources
4
Our benchmarks
0
Price per million tokens
$0.09 in · $0.30 out
OpenRouter list price, 1 Oct 2026 · 262,144-token context

Reported by others

Reported by Arena (formerly LMArena)
1,445
Overall · rank 53 of 177
Unit
Arena rating, higher is better
Range
1,438 to 1,452
Sample
6076 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not an error rate: claims that cannot be checked on the web are skipped, and preference still carries most of the weight.
Reported by Arena (formerly LMArena)
1,444
Business, management and finance · rank 37 of 402
Unit
Arena rating, higher is better
Range
1,427 to 1,461
Sample
1115 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
Reported by Arena (formerly LMArena)
1,399
Creative writing · rank 72 of 407
Unit
Arena rating, higher is better
Range
1,381 to 1,418
Sample
994 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
Reported by Arena (formerly LMArena)
1,475
Expert prompts · rank 22 of 359
Unit
Arena rating, higher is better
Range
1,450 to 1,501
Sample
456 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
Reported by Arena (formerly LMArena)
1,437
Instruction following · rank 48 of 409
Unit
Arena rating, higher is better
Range
1,424 to 1,451
Sample
1738 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
Reported by Arena (formerly LMArena)
1,438
Overall · rank 76 of 409
Unit
Arena rating, higher is better
Range
1,430 to 1,445
Sample
6078 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
Reported by Arena (formerly LMArena)
1,406
Writing, literature and language · rank 78 of 408
Unit
Arena rating, higher is better
Range
1,391 to 1,421
Sample
1394 votes
Configuration
Gemma 4 26B A4B
Measured
25 Sep 2026
Not shown
Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
Reported by Artificial Analysis
39.3
Artificial Analysis Coding Index · rank 114 of 151
Unit
index score, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not your codebase or tools.
Reported by Artificial Analysis
16.7
Artificial Analysis Intelligence Index · rank 205 of 314
Unit
index score, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not business work, and a blend: read the parts for any one task.
Reported by Artificial Analysis
13.1
Artificial Analysis Intelligence Index · rank 237 of 314
Unit
index score, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not business work, and a blend: read the parts for any one task.
Reported by Artificial Analysis
79.2%
GPQA Diamond · rank 160 of 276
Unit
% of questions, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not applied work; multiple-choice science questions.
Reported by Artificial Analysis
71.4%
GPQA Diamond · rank 208 of 276
Unit
% of questions, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not applied work; multiple-choice science questions.
Reported by Artificial Analysis
19.3%
Humanity's Last Exam · rank 170 of 314
Unit
% of questions, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not everyday work; academic questions at the edge of expertise.
Reported by Artificial Analysis
11.5%
Humanity's Last Exam · rank 216 of 314
Unit
% of questions, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not everyday work; academic questions at the edge of expertise.
Reported by Artificial Analysis
72.4%
IFBench · rank 41 of 215
Unit
% of instructions, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not judgement about what an instruction meant.
Reported by Artificial Analysis
45.4%
IFBench · rank 138 of 215
Unit
% of instructions, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not judgement about what an instruction meant.
Reported by Artificial Analysis
65.7%
Long-context reasoning (AA-LCR) · rank 191 of 309
Unit
% of questions, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not retrieval over your own document store.
Reported by Artificial Analysis
42.3%
Long-context reasoning (AA-LCR) · rank 253 of 309
Unit
% of questions, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not retrieval over your own document store.
Reported by Artificial Analysis
40.0%
SciCode · rank 130 of 153
Unit
% of problems, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not general software engineering.
Reported by Artificial Analysis
39.0%
Terminal-Bench 2.1 · rank 115 of 151
Unit
% of tasks, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not other harnesses or tools; superseded by 4.0 for newer models.
Reported by Artificial Analysis
13.6%
Terminal-Bench Hard · rank 158 of 213
Unit
% of tasks, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not other harnesses; Artificial Analysis no longer runs it on new models.
Reported by Artificial Analysis
25.0%
Terminal-Bench Hard · rank 111 of 213
Unit
% of tasks, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not other harnesses; Artificial Analysis no longer runs it on new models.
Reported by Artificial Analysis
12.0%
Τ-bench banking · rank 110 of 142
Unit
% of tasks, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not your policies or systems; a simulated customer.
Reported by Artificial Analysis
43.6%
Τ²-bench telecom · rank 146 of 214
Unit
% of tasks, higher is better
Configuration
Gemma 4 26B A4B
Measured
1 Oct 2026
Not shown
Not your policies or systems; no longer run on new models.
Reported by Artificial Analysis
40.4%
Τ²-bench telecom · rank 153 of 214
Unit
% of tasks, higher is better
Configuration
Gemma 4 26B A4B (off)
Measured
1 Oct 2026
Not shown
Not your policies or systems; no longer run on new models.
Reported by UGI Leaderboard
8.0%
Requested-length error · rank 55 of 370
Unit
% off the requested word count, lower is better
Configuration
Gemma 4 26B A4B
Measured
12 Apr 2026
Not shown
Not other format limits such as character counts or bullet counts.
Reported by UGI Leaderboard
8.0%
Requested-length error · rank 55 of 370
Unit
% off the requested word count, lower is better
Configuration
Gemma 4 26B A4B (channel-thought-prefill reasoning)
Measured
12 Apr 2026
Not shown
Not other format limits such as character counts or bullet counts.
Reported by UGI Leaderboard
0.34
Style adherence · rank 215 of 370
Unit
score from 0 to 1, higher is better
Configuration
Gemma 4 26B A4B
Measured
12 Apr 2026
Not shown
Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.
Reported by UGI Leaderboard
0.35
Style adherence · rank 166 of 370
Unit
score from 0 to 1, higher is better
Configuration
Gemma 4 26B A4B (channel-thought-prefill reasoning)
Measured
12 Apr 2026
Not shown
Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.
Reported by UGI Leaderboard
41.6
Writing score · rank 204 of 370
Unit
score out of 100, higher is better
Configuration
Gemma 4 26B A4B
Measured
12 Apr 2026
Not shown
Not business copy quality: UGI's prompts are private and lean towards creative writing, and models that often refuse get no score.
Reported by UGI Leaderboard
43.8
Writing score · rank 187 of 370
Unit
score out of 100, higher is better
Configuration
Gemma 4 26B A4B (channel-thought-prefill reasoning)
Measured
12 Apr 2026
Not shown
Not business copy quality: UGI's prompts are private and lean towards creative writing, and models that often refuse get no score.
Reported by Vectara Hallucination Leaderboard
99.8%
Answer rate · rank 34 of 108
Unit
% of documents, higher is better
Configuration
Gemma 4 26B A4B
Measured
22 Sep 2026
Not shown
Not a quality score: a low rate usually means content filters were triggered, and hallucination rates are measured on answered documents only.
Reported by Vectara Hallucination Leaderboard
5.2%
Hallucination rate · rank 13 of 108
Unit
% of summaries, lower is better
Configuration
Gemma 4 26B A4B
Measured
22 Sep 2026
Not shown
Not errors in open questions or other tasks: only summarisation, judged by Vectara's own model (HHEM-2.3), not by people, on news-style documents rather than your data.

Compare Gemma 4 26B A4B with