Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Gemma 4 E4B
~47
0/8 categoriesQwen3.5 397B
68
Winner · 4/8 categoriesGemma 4 E4B· Qwen3.5 397B
Pick Qwen3.5 397B if you want the stronger benchmark profile. Gemma 4 E4B only becomes the better choice if you want the stronger reasoning-first profile.
Qwen3.5 397B is clearly ahead on the aggregate, 68 to 47. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Qwen3.5 397B's sharpest advantage is in reasoning, where it averages 69.7 against 25.4. The single biggest benchmark swing on the page is BBH, 33.1% to 82%.
Gemma 4 E4B is the reasoning model in the pair, while Qwen3.5 397B 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.
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 E4B | Qwen3.5 397B |
|---|---|---|
| Agentic | ||
| Terminal-Bench 2.0 | — | 52.5% |
| BrowseComp | — | 62% |
| OSWorld-Verified | — | 62.2% |
| Claw-Eval | — | 48.1% |
| QwenClawBench | — | 51.8% |
| QwenWebBench | — | 1162 |
| TAU3-Bench | — | 68.4% |
| DeepPlanning | — | 37.6% |
| Toolathlon | — | 36.3% |
| MCP Atlas | — | 46.1% |
| MCP-Tasks | — | 74.2% |
| WideResearch | — | 74.0% |
| CodingQwen3.5 397B wins | ||
| LiveCodeBench | 52% | 39% |
| HumanEval | — | 75% |
| SWE-bench Verified | — | 76.2% |
| LiveCodeBench v6 | — | 83.6% |
| SWE-bench Pro | — | 50.9% |
| SWE Multilingual | — | 69.3% |
| NL2Repo | — | 32.2% |
| Multimodal & GroundedQwen3.5 397B wins | ||
| MMMU-Pro | 52.6% | 79% |
| OfficeQA Pro | — | 68% |
| RealWorldQA | — | 83.9% |
| Video-MME (w/o subtitle) | — | 84.2% |
| MathVision | — | 88.6% |
| We-Math | — | 87.9% |
| DynaMath | — | 86.3% |
| MStar | — | 83.8% |
| SimpleVQA | — | 67.1% |
| ChatCVQA | — | 80.8% |
| AI2D_TEST | — | 93.9% |
| CountBench | — | 97.2% |
| RefCOCO (avg) | — | 92.3% |
| ODINW13 | — | 47.0% |
| MLVU (M-Avg) | — | 86.7% |
| ScreenSpot Pro | — | 65.6% |
| ReasoningQwen3.5 397B wins | ||
| BBH | 33.1% | 82% |
| MRCRv2 | 25.4% | 71% |
| MuSR | — | 78% |
| LongBench v2 | — | 63.2% |
| AI-Needle | — | 68.7% |
| KnowledgeQwen3.5 397B wins | ||
| GPQA | 58.6% | 88.4% |
| MMLU-Pro | 69.4% | 87.8% |
| MMLU | — | 83% |
| SuperGPQA | — | 70.4% |
| MMLU-Redux | — | 94.9% |
| C-Eval | — | 93% |
| HLE | — | 28.7% |
| FrontierScience | — | 71% |
| SimpleQA | — | 80% |
| Instruction Following | ||
| IFEval | — | 92.6% |
| IFBench | — | 76.5% |
| Multilingual | ||
| MGSM | — | 82% |
| MMLU-ProX | — | 84.7% |
| NOVA-63 | — | 59.1% |
| INCLUDE | — | 85.6% |
| PolyMath | — | 73.3% |
| VWT2k-lite | — | 78.9% |
| MAXIFE | — | 88.2% |
| Mathematics | ||
| AIME 2023 | — | 83% |
| AIME 2024 | — | 85% |
| AIME 2025 | — | 84% |
| AIME26 | — | 93.3% |
| HMMT Feb 2023 | — | 79% |
| HMMT Feb 2024 | — | 81% |
| HMMT Feb 2025 | — | 80% |
| HMMT Feb 2025 | — | 94.8% |
| HMMT Nov 2025 | — | 92.7% |
| HMMT Feb 2026 | — | 87.9% |
| MMAnswerBench | — | 80.9% |
| BRUMO 2025 | — | 82% |
| MATH-500 | — | 81% |
Qwen3.5 397B is ahead overall, 68 to 47. The biggest single separator in this matchup is BBH, where the scores are 33.1% and 82%.
Qwen3.5 397B has the edge for knowledge tasks in this comparison, averaging 68.2 versus 65.6. Inside this category, GPQA is the benchmark that creates the most daylight between them.
Qwen3.5 397B has the edge for coding in this comparison, averaging 52.2 versus 52. Inside this category, LiveCodeBench is the benchmark that creates the most daylight between them.
Qwen3.5 397B has the edge for reasoning in this comparison, averaging 69.7 versus 25.4. Inside this category, BBH is the benchmark that creates the most daylight between them.
Qwen3.5 397B has the edge for multimodal and grounded tasks in this comparison, averaging 74.1 versus 52.6. 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.