Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Grok 4
68
2/8 categoriesQwen3.5-27B
71
Winner · 5/8 categoriesGrok 4· Qwen3.5-27B
Pick Qwen3.5-27B if you want the stronger benchmark profile. Grok 4 only becomes the better choice if agentic is the priority or you would rather avoid the extra latency and token burn of a reasoning model.
Qwen3.5-27B has the cleaner overall profile here, landing at 71 versus 68. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
Qwen3.5-27B's sharpest advantage is in knowledge, where it averages 80.6 against 65.3. The single biggest benchmark swing on the page is SuperGPQA, 84% to 65.6%. Grok 4 does hit back in agentic, so the answer changes if that is the part of the workload you care about most.
Qwen3.5-27B is the reasoning model in the pair, while Grok 4 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. Qwen3.5-27B gives you the larger context window at 262K, compared with 128K for Grok 4.
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 | Grok 4 | Qwen3.5-27B |
|---|---|---|
| AgenticGrok 4 wins | ||
| Terminal-Bench 2.0 | 57% | 41.6% |
| BrowseComp | 63% | 61% |
| OSWorld-Verified | 56% | 56.2% |
| tau2-bench | — | 79% |
| CodingQwen3.5-27B wins | ||
| HumanEval | 79% | — |
| SWE-bench Verified | 73% | 72.4% |
| LiveCodeBench | 79.4% | 80.7% |
| SWE-bench Pro | 48% | — |
| React Native Evals | 70.1% | — |
| Multimodal & GroundedGrok 4 wins | ||
| MMMU-Pro | 80% | 75% |
| OfficeQA Pro | 76% | — |
| ReasoningQwen3.5-27B wins | ||
| MuSR | 81% | — |
| BBH | 83% | — |
| LongBench v2 | 72% | 60.6% |
| MRCRv2 | 71% | — |
| ARC-AGI-2 | 16% | — |
| KnowledgeQwen3.5-27B wins | ||
| MMLU | 87% | — |
| GPQA | 86% | 85.5% |
| SuperGPQA | 84% | 65.6% |
| MMLU-Pro | 77% | 86.1% |
| HLE | 16% | — |
| FrontierScience | 75% | — |
| SimpleQA | 83% | — |
| Instruction FollowingQwen3.5-27B wins | ||
| IFEval | 82% | 95% |
| MultilingualQwen3.5-27B wins | ||
| MGSM | 84% | — |
| MMLU-ProX | 79% | 82.2% |
| Mathematics | ||
| AIME 2023 | 87% | — |
| AIME 2024 | 89% | — |
| AIME 2025 | 88% | — |
| HMMT Feb 2023 | 84% | — |
| HMMT Feb 2024 | 86% | — |
| HMMT Feb 2025 | 85% | — |
| BRUMO 2025 | 87% | — |
| MATH-500 | 83% | — |
Qwen3.5-27B is ahead overall, 71 to 68. The biggest single separator in this matchup is SuperGPQA, where the scores are 84% and 65.6%.
Qwen3.5-27B has the edge for knowledge tasks in this comparison, averaging 80.6 versus 65.3. Inside this category, SuperGPQA is the benchmark that creates the most daylight between them.
Qwen3.5-27B has the edge for coding in this comparison, averaging 77.6 versus 65.8. Inside this category, LiveCodeBench is the benchmark that creates the most daylight between them.
Qwen3.5-27B has the edge for reasoning in this comparison, averaging 60.6 versus 59.6. Inside this category, LongBench v2 is the benchmark that creates the most daylight between them.
Grok 4 has the edge for agentic tasks in this comparison, averaging 58.1 versus 51.6. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.
Grok 4 has the edge for multimodal and grounded tasks in this comparison, averaging 78.2 versus 75. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.
Qwen3.5-27B has the edge for instruction following in this comparison, averaging 95 versus 82. Inside this category, IFEval is the benchmark that creates the most daylight between them.
Qwen3.5-27B has the edge for multilingual tasks in this comparison, averaging 82.2 versus 80.8. Inside this category, MMLU-ProX 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.