Google model product
Gemma 4 26B-A4B is Google DeepMind's Mixture-of-Experts multimodal model with 26 billion total parameters and 3.8 billion activated parameters per token, with a 256K context window. Ranks #6 among open models on Arena AI, outcompeting models 20x its size. Achieves an estimated LMArena text score of 1441. Supports image and text inputs.
Updated Aug 10, 2026. Default version: Gemma 4 26B-A4B
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version Gemma 4 26B-A4B.
| BIG-Bench Extra Hard | 0.6 | 2 | 11 | 90.0% | C | |
| MedXpertQA | 0.6 | 6 | 12 | 54.5% | C | |
| t2-bench | 0.9 | 6 | 23 | 77.3% | C | |
| MRCR v2 (8-needle) | 0.4 | 8 | 21 | 65.0% | C | |
| AIME 2026 | 0.9 | 12 | 17 | 31.3% | C | |
| MathVision | 0.8 | 12 | 32 | 64.5% | C | |
| LiveCodeBench v6 | 0.8 | 23 | 53 | 57.7% | C | |
| MMMLU | 0.9 | 24 | 49 | 52.1% | C | |
| MMLU-Pro | 0.8 | 34 | 129 | 74.2% | C | |
| MMMU-Pro | 0.7 | 35 | 65 | 46.9% | C | |
| Humanity's Last Exam | 0.2 | 64 | 92 | 30.8% | C | |
| GPQA | 0.8 | 73 | 233 | 69.0% | C |
Preference and agent-evaluation signals from published Arena datasets.
| vision | captioning | 4 | 1300.7 | 43 | N/A | |
| vision style control | captioning | 5 | 1247.8 | 43 | N/A | |
| vision | humor | 22 | 1278.6 | 877 | N/A | |
| vision | homework | 29 | 1292.9 | 3,614 | N/A | |
| vision style control | humor | 29 | 1251.9 | 877 | N/A | |
| vision style control | homework | 31 | 1284.9 | 3,614 | N/A | |
| text | math | 33 | 1467.2 | 370 | N/A | |
| vision | chinese | 33 | 1315.7 | 1,350 | N/A | |
| vision | creative writing vision | 34 | 1260.9 | 1,328 | N/A | |
| vision style control | chinese | 36 | 1294.8 | 1,350 | N/A | |
| text | industry mathematical | 37 | 1470.0 | 292 | N/A | |
| text style control | industry mathematical | 37 | 1475.3 | 292 | N/A | |
| vision | ocr | 39 | 1266.4 | 15,709 | N/A | |
| text | french | 40 | 1466.2 | 216 | N/A | |
| vision style control | entity recognition | 40 | 1152.8 | 205 | N/A | |
| text style control | math | 41 | 1467.6 | 370 | N/A | |
| vision | overall | 41 | 1257.8 | 22,246 | N/A | |
| vision | entity recognition | 41 | 1182.1 | 205 | N/A | |
| vision style control | creative writing vision | 44 | 1235.3 | 1,328 | N/A | |
| text style control | french | 45 | 1477.5 | 216 | N/A | |
| vision style control | ocr | 45 | 1255.8 | 15,709 | N/A | |
| text | industry legal and government | 46 | 1456.2 | 382 | N/A | |
| vision | diagram | 46 | 1263.3 | 5,777 | N/A | |
| vision | english | 47 | 1254.4 | 9,046 | N/A | |
| vision style control | overall | 48 | 1239.6 | 22,246 | N/A | |
| text | industry business and management and financial operations | 50 | 1434.7 | 1,042 | N/A | |
| text | russian | 50 | 1443.3 | 678 | N/A | |
| vision style control | diagram | 51 | 1260.6 | 5,777 | N/A | |
| vision style control | english | 51 | 1236.8 | 9,046 | N/A | |
| text factuality | instruction following | 55 | 1442.7 | 1,588 | N/A | |
| text style control | industry legal and government | 56 | 1462.4 | 382 | N/A | |
| text factuality | exclude ties | 59 | 1447.3 | 3,943 | N/A | |
| text style control | industry business and management and financial operations | 60 | 1449.5 | 1,042 | N/A | |
| text | chinese | 61 | 1477.8 | 389 | N/A | |
| text | instruction following | 62 | 1422.5 | 1,590 | N/A | |
| text style control | instruction following | 62 | 1439.8 | 1,590 | N/A | |
| text | expert | 63 | 1448.3 | 405 | N/A | |
| text factuality | overall | 63 | 1445.1 | 5,753 | N/A | |
| text factuality | english | 63 | 1454.5 | 2,492 | N/A | |
| text style control | expert | 63 | 1475.3 | 405 | N/A | |
| text | industry entertainment and sports and media | 65 | 1406.2 | 1,037 | N/A | |
| text factuality | non english | 65 | 1433.2 | 3,261 | N/A | |
| text style control | russian | 65 | 1446.8 | 678 | N/A | |
| text | non english | 67 | 1423.7 | 3,262 | N/A | |
| text factuality | hard prompts english | 67 | 1473.3 | 1,469 | N/A | |
| webdev | webdev-react | 68 | 1358.5 | 1,220 | N/A | |
| text | english | 69 | 1442.5 | 2,493 | N/A | |
| text style control | hard prompts english | 69 | 1469.9 | 1,470 | N/A | |
| text | exclude ties | 70 | 1429.8 | 3,945 | N/A | |
| text style control | longer query | 71 | 1448.5 | 1,553 | N/A | |
| webdev | webdev-html | 71 | 1375.0 | 190 | N/A | |
| text | overall | 72 | 1434.7 | 5,755 | N/A | |
| text style control | hard prompts | 72 | 1462.3 | 3,245 | N/A | |
| text | creative writing | 73 | 1404.5 | 939 | N/A | |
| text | longer query | 73 | 1429.4 | 1,553 | N/A | |
| text factuality | hard prompts | 73 | 1463.0 | 3,243 | N/A | |
| text style control | industry entertainment and sports and media | 73 | 1410.3 | 1,037 | N/A | |
| webdev | overall | 73 | 1366.5 | 1,415 | N/A | |
| webdev | webdev | 73 | 1366.5 | 1,415 | N/A | |
| text | hard prompts english | 74 | 1446.1 | 1,470 | N/A | |
| text | industry software and it services | 74 | 1453.1 | 2,077 | N/A | |
| text | multi turn | 74 | 1438.9 | 1,082 | N/A | |
| text factuality | industry software and it services | 74 | 1476.8 | 2,077 | N/A | |
| text | hard prompts | 76 | 1439.9 | 3,245 | N/A | |
| text style control | english | 76 | 1449.2 | 2,493 | N/A | |
| text style control | industry software and it services | 76 | 1474.8 | 2,077 | N/A | |
| text | industry writing and literature and language | 77 | 1409.1 | 1,303 | N/A | |
| text factuality | longer query | 77 | 1448.2 | 1,553 | N/A | |
| text factuality | coding | 78 | 1484.4 | 1,357 | N/A | |
| text style control | overall | 78 | 1438.2 | 5,755 | N/A | |
| text style control | exclude ties | 78 | 1435.3 | 3,945 | N/A | |
| text | industry life and physical and social science | 80 | 1442.3 | 865 | N/A | |
| text style control | multi turn | 80 | 1446.1 | 1,082 | N/A | |
| text style control | non english | 82 | 1424.6 | 3,262 | N/A | |
| text style control | chinese | 85 | 1475.0 | 389 | N/A | |
| text style control | coding | 85 | 1480.6 | 1,357 | N/A | |
| text style control | creative writing | 86 | 1403.0 | 939 | N/A | |
| text style control | industry writing and literature and language | 86 | 1409.7 | 1,303 | N/A | |
| text | coding | 88 | 1444.6 | 1,357 | N/A | |
| text style control | industry life and physical and social science | 89 | 1450.7 | 865 | N/A | |
| text | industry medicine and healthcare | 108 | 1423.0 | 364 | N/A | |
| text style control | industry medicine and healthcare | 126 | 1431.4 | 364 | N/A |
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
| EmpirioLabs AI | gemma-4-26b-a4b | global | $0.05 | $0.29 | 262.1K | |
| LLM Gateway | gemma-4-26b-a4b-it | global | $0.07 | $0.34 | 262.1K | |
| Cortecs | gemma-4-26b-a4b-it | global | $0.111 | $0.557 | 262K | |
| Charm Hyper | gemma-4-26b-a4b-it | global | $0.124 | $0.42 | 256K | |
| Scaleway | gemma-4-26b-a4b-it | global | $0.25 | $0.50 | 256K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| Gemma 4 26B-A4B | 67.2 | 25.2B | 262.1K | 32.8K | Yes | Apache 2.0 |
A concise description based on the published model registry.
Gemma 4 26B-A4B is Google DeepMind's Mixture-of-Experts multimodal model with 26 billion total parameters and 3.8 billion activated parameters per token, with a 256K context window. Ranks #6 among open models on Arena AI, outcompeting models 20x its size. Achieves an estimated LMArena text score of 1441. Supports image and text inputs.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about Gemma 4 26B A4B.
Gemma 4 26B A4B's default version was released on Apr 2, 2026.
No official standard PAYG price is currently available for Gemma 4 26B A4B. The lowest tracked third-party offer starts at $0.05 input and $0.29 output via EmpirioLabs AI.
Gemma 4 26B A4B is published under Google in the model registry.
The default version has a 262.1K token context window.
Yes. The default version is marked as open weight under Apache 2.0.
5 published provider offerings are linked to the default version.
Nearby ranked alternatives include Claude Opus 4.6, GPT-5.2, MiMo V2.5 Pro.