- MMLU Pro 82.6
- AIME 2026 no tools 88.3
- LiveCodeBench v6 77.1
- Codeforces ELO 1718
- GPQA Diamond 82.3
- Tau2 (average over 3) 68.2
- Best of HLE no tools 8.7 / HLE with search 17.2 17.2
- BigBench Extra Hard 64.8
- MMMLU 86.3
- MMMU Pro 73.8
- OmniDocBench 1.5 (average edit distance, lower is better) 0.149
- MATH-Vision 82.4
- MedXPertQA MM 58.1
- MRCR v2 8 needle 128k (average) 44.1
Gemma 4 26B A4B
Gemma 4 26B A4B places in the top three on 3 benchmarks.
Google DeepMind Released 2 Apr 2026 57
Benchmarks with a score
0
Ranked first, out of five or more models
3
In the top three
Google DeepMind Apr 2026 Gemma 4 26B A4B
Open weights
141.9
#113 of 271
Epoch Capabilities Index
- IFBench 72% #35 /70
- MMMU-Pro 73.8% #38 /72
- SciCode 40% #30 /47
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| IFEval | 98.5% | #2 of 126 | Model card | |
| Vectara Hallucination Leaderboard (2025 dataset) | 5.20% | #14 of 104 | Official leaderboard | |
| FACTS Grounding v2 | 77.8% | #3 of 15 | Official leaderboard | |
| SEA-HELM | 74.8 | #14 of 62 | Official leaderboard | |
| LMArena (Elo) | 1,438 | #80 of 333 | Official leaderboard | |
| BIG-Bench Extra Hard | 64.8% | #3 of 11 | Model card | |
| EQ-Bench 3 | 1,455 | #27 of 78 | Official leaderboard | |
| MMLU-Pro | 82.6% | #53 of 151 | Model card | |
| FACTS Benchmark Suite | 54% | #15 of 40 | Official leaderboard | |
| MMMLU | 86.3% | #24 of 62 | Model card | |
| Wer wird Millionär? LLM Benchmark | 10667.0 | #34 of 76 | Official leaderboard | |
| IFBench | 72% | #35 of 70 | Model card | |
| RULER | 89.8% | #20 of 38 | Model card | |
| Creative Writing v3 | 65.8 | #66 of 120 | Official leaderboard | |
| Judgemark v4 | 53.0 | #25 of 40 | Official leaderboard | |
| Global MMLU Lite | 85.7% | #28 of 43 | Official leaderboard | |
| GraphWalks | 72.6% | #14 of 21 | Model card | |
| BullshitBench V2 | 21% | #84 of 122 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 50.7 | #82 of 119 | Official leaderboard | |
| Humanity's Last Exam | 17.2% | #112 of 147 | Model card | |
| BullshitBench | 14.6% | #89 of 107 | Official leaderboard | |
| SimpleQA Verified | 9.05% | #49 of 56 | Official leaderboard | |
| MRCR v2 | 44.1% | #24 of 27 | Model card | |
| FACTS Multimodal | 23.9% | #35 of 39 | Official leaderboard | |
| FACTS Search | 56.7% | #28 of 31 | Official leaderboard | |
| FACTS Parametric | 5.69% | #41 of 45 | Official leaderboard | |
| FACTS Search V2 | 60.3% | #21 of 21 | Official leaderboard | |
| LOFT | 66.3% | #6 of 6 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 82.2% | #85 of 202 | Epoch AI | |
| GPQA Diamond | 73.2% | #103 of 219 | Epoch AI | |
| DTBench | 58.2% | #105 of 178 | Official leaderboard | |
| AIME 2026 | 88.3% | #29 of 34 | Model card | |
| CritPt | 0% | #155 of 160 | Epoch AI |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench v6 | 77.1% | #36 of 76 | Model card | |
| ALE-Bench | 927 | #59 of 121 | Epoch AI | |
| SciCode | 40% | #30 of 47 | Model card | |
| WebDev Arena | 1,361 | #79 of 105 | Epoch AI | |
| Code Arena | 1,359 | #94 of 124 | Official leaderboard | |
| WeirdML | 35.2% | #100 of 130 | Epoch AI | |
| StructEval | 68.9% | #22 of 28 | Official leaderboard | |
| Codeforces rating | 1,718 | #33 of 41 | Model card | |
| ScarfBench | 24.4% | #10 of 10 | Official leaderboard |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PinchBench | 74.1% | #13 of 48 | Official leaderboard | |
| SnakeBench | 24 | #94 of 183 | Official leaderboard | |
| PinchBench 2.0 | 56.4% | #43 of 57 | Official leaderboard | |
| tau2-bench | 68.2% | #55 of 69 | Model card | |
| Terminal-Bench Hard | 14% | #9 of 11 | Model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MedXpertQA-MM | 58.1% | #8 of 12 | Model card | |
| Physical-Engineering's Last Exam | 0.04 | #16 of 18 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| InfographicVQA | 89.3% | #5 of 22 | Model card | |
| MATH-Vision (MathVision) | 82.4% | #16 of 42 | Model card | |
| Vision Arena | 1,242 | #55 of 136 | Official leaderboard | |
| MMMU-Pro | 73.8% | #38 of 72 | Model card | |
| OmniDocBench v1.5 | 0.15 | #8 of 13 | Model card |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMCA | 29.7% | #81 of 139 | Epoch AI | |
| Conceptual Reasoning Index | 40.3 | #97 of 149 | Official leaderboard | |
| ACCoRD | 54.3 | #107 of 150 | Official leaderboard |
Google DeepMind's models over time
Google DeepMind's models in color, against every other model on the Epoch Capabilities Index, by release date.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it. 10 of them aren't in the launch post.
Gemma 4 Technical Report (arXiv 2607.02770v2) 24 Jul 2026
Capabilities
- LMArena (Elo) 1438
- MMLU-Pro 82.6
- AIME 2026 88.3
- LiveCodeBench v6 77.1
- Codeforces rating 1718
- SciCode 40
- GPQA Diamond 82.3
- BIG-Bench Extra Hard 64.8
- Humanity's Last Exam 8.7
- IFBench 72
- IFEval 98.5
- MMMLU 86.3
- MRCR v2 44.1
- Terminal-Bench Hard 14
- tau2-bench
- MMMU-Pro 73.8
- MATH-Vision (MathVision) 82.4
- MedXpertQA-MM 58.1
- InfographicVQA 89.3
- OmniDocBench v1.5 0.149
- RULER 89.8
- LOFT 66.3
- GraphWalks 72.6
- MTOB