Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
DeepSeek-R1
45
1/8 categoriesGemma 4 E4B
~47
Winner · 3/8 categoriesDeepSeek-R1· Gemma 4 E4B
Pick Gemma 4 E4B if you want the stronger benchmark profile. DeepSeek-R1 only becomes the better choice if reasoning is the priority.
Gemma 4 E4B has the cleaner overall profile here, landing at 47 versus 45. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
Gemma 4 E4B's sharpest advantage is in coding, where it averages 52 against 28.3. The single biggest benchmark swing on the page is LiveCodeBench, 19% to 52%. DeepSeek-R1 does hit back in reasoning, so the answer changes if that is the part of the workload you care about most.
DeepSeek-R1 is also the more expensive model on tokens at $0.55 input / $2.19 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Gemma 4 E4B. That is roughly Infinityx on output cost alone.
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 | DeepSeek-R1 | Gemma 4 E4B |
|---|---|---|
| Agentic | ||
| Terminal-Bench 2.0 | 42% | — |
| BrowseComp | 49% | — |
| OSWorld-Verified | 44% | — |
| CodingGemma 4 E4B wins | ||
| HumanEval | 92% | — |
| SWE-bench Verified | 49.2% | — |
| LiveCodeBench | 19% | 52% |
| SWE-bench Pro | 25% | — |
| Multimodal & GroundedGemma 4 E4B wins | ||
| MMMU-Pro | 43% | 52.6% |
| OfficeQA Pro | 53% | — |
| ReasoningDeepSeek-R1 wins | ||
| MuSR | 40% | — |
| BBH | 66% | 33.1% |
| LongBench v2 | 58% | — |
| MRCRv2 | 57% | 25.4% |
| ARC-AGI-2 | 1.3% | — |
| KnowledgeGemma 4 E4B wins | ||
| MMLU | 90.8% | — |
| GPQA | 71.5% | 58.6% |
| SuperGPQA | 41% | — |
| MMLU-Pro | 84% | 69.4% |
| HLE | 14% | — |
| FrontierScience | 44% | — |
| SimpleQA | 30.1% | — |
| Instruction Following | ||
| IFEval | 83.3% | — |
| Multilingual | ||
| MGSM | 61% | — |
| MMLU-ProX | 60% | — |
| Mathematics | ||
| AIME 2023 | 44% | — |
| AIME 2024 | 79.8% | — |
| AIME 2025 | 45% | — |
| HMMT Feb 2023 | 40% | — |
| HMMT Feb 2024 | 42% | — |
| HMMT Feb 2025 | 41% | — |
| BRUMO 2025 | 43% | — |
| MATH-500 | 97.3% | — |
Gemma 4 E4B is ahead overall, 47 to 45. The biggest single separator in this matchup is LiveCodeBench, where the scores are 19% and 52%.
Gemma 4 E4B has the edge for knowledge tasks in this comparison, averaging 65.6 versus 47. Inside this category, MMLU-Pro is the benchmark that creates the most daylight between them.
Gemma 4 E4B has the edge for coding in this comparison, averaging 52 versus 28.3. Inside this category, LiveCodeBench is the benchmark that creates the most daylight between them.
DeepSeek-R1 has the edge for reasoning in this comparison, averaging 40 versus 25.4. Inside this category, BBH is the benchmark that creates the most daylight between them.
Gemma 4 E4B has the edge for multimodal and grounded tasks in this comparison, averaging 52.6 versus 47.5. 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.