Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Gemma 4 31B
73
Winner · 4/8 categoriesGPT-OSS 120B
50
0/8 categoriesGemma 4 31B· GPT-OSS 120B
Pick Gemma 4 31B if you want the stronger benchmark profile. GPT-OSS 120B only becomes the better choice if you would rather avoid the extra latency and token burn of a reasoning model.
Gemma 4 31B is clearly ahead on the aggregate, 73 to 50. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Gemma 4 31B's sharpest advantage is in coding, where it averages 80 against 30. The single biggest benchmark swing on the page is LiveCodeBench, 80% to 25%.
Gemma 4 31B is the reasoning model in the pair, while GPT-OSS 120B is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. Gemma 4 31B gives you the larger context window at 256K, compared with 128K for GPT-OSS 120B.
BenchLM keeps the benchmark table and the operator tradeoffs on the same page so a better score does not hide a materially slower, pricier, or smaller-context model.
Runtime metrics show N/A when BenchLM does not have a sourced snapshot for that exact model. The scoring rules and freshness policy are documented on the methodology page.
| Benchmark | Gemma 4 31B | GPT-OSS 120B |
|---|---|---|
| Agentic | ||
| Terminal-Bench 2.0 | — | 43% |
| BrowseComp | — | 50% |
| OSWorld-Verified | — | 43% |
| CodingGemma 4 31B wins | ||
| LiveCodeBench | 80% | 25% |
| HumanEval | — | 43% |
| SWE-bench Verified | — | 29% |
| SWE-bench Pro | — | 31% |
| SWE-Rebench | — | 33.3% |
| React Native Evals | — | 66.4% |
| Multimodal & GroundedGemma 4 31B wins | ||
| MMMU-Pro | 76.9% | 42% |
| OfficeQA Pro | — | 57% |
| ReasoningGemma 4 31B wins | ||
| BBH | 74.4% | 73% |
| MRCRv2 | 66.4% | 59% |
| MuSR | — | 47% |
| LongBench v2 | — | 58% |
| KnowledgeGemma 4 31B wins | ||
| GPQA | 84.3% | 80.9% |
| MMLU-Pro | 85.2% | 90% |
| HLE | 26.5% | 5% |
| HLE w/o tools | 19.5% | — |
| MMLU | — | 90% |
| SuperGPQA | — | 48% |
| FrontierScience | — | 49% |
| SimpleQA | — | 49% |
| Instruction Following | ||
| IFEval | — | 79% |
| Multilingual | ||
| MGSM | — | 72% |
| MMLU-ProX | — | 70% |
| Mathematics | ||
| AIME 2023 | — | 51% |
| AIME 2024 | — | 53% |
| AIME 2025 | — | 52% |
| HMMT Feb 2023 | — | 47% |
| HMMT Feb 2024 | — | 49% |
| HMMT Feb 2025 | — | 48% |
| BRUMO 2025 | — | 50% |
| MATH-500 | — | 71% |
Gemma 4 31B is ahead overall, 73 to 50. The biggest single separator in this matchup is LiveCodeBench, where the scores are 80% and 25%.
Gemma 4 31B has the edge for knowledge tasks in this comparison, averaging 61.3 versus 51.6. Inside this category, HLE is the benchmark that creates the most daylight between them.
Gemma 4 31B has the edge for coding in this comparison, averaging 80 versus 30. Inside this category, LiveCodeBench is the benchmark that creates the most daylight between them.
Gemma 4 31B has the edge for reasoning in this comparison, averaging 66.4 versus 55.4. Inside this category, MRCRv2 is the benchmark that creates the most daylight between them.
Gemma 4 31B has the edge for multimodal and grounded tasks in this comparison, averaging 76.9 versus 48.8. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.
Get notified when new models drop, benchmark scores change, or the leaderboard shifts. One email per week.
Free. No spam. Unsubscribe anytime. We only store derived location metadata for consent routing.