Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Phi-4
49
0/8 categoriesQwen3.5-122B-A10B
71
Winner · 6/8 categoriesPhi-4· Qwen3.5-122B-A10B
Pick Qwen3.5-122B-A10B if you want the stronger benchmark profile. Phi-4 only becomes the better choice if you would rather avoid the extra latency and token burn of a reasoning model.
Qwen3.5-122B-A10B is clearly ahead on the aggregate, 71 to 49. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Qwen3.5-122B-A10B's sharpest advantage is in multimodal & grounded, where it averages 76.9 against 46.8. The single biggest benchmark swing on the page is GPQA, 56.1% to 86.6%.
Qwen3.5-122B-A10B is the reasoning model in the pair, while Phi-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. Qwen3.5-122B-A10B gives you the larger context window at 262K, compared with 16K for Phi-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 | Phi-4 | Qwen3.5-122B-A10B |
|---|---|---|
| AgenticQwen3.5-122B-A10B wins | ||
| Terminal-Bench 2.0 | 44% | 49.4% |
| BrowseComp | 35% | 63.8% |
| OSWorld-Verified | 34% | 58% |
| tau2-bench | — | 79.5% |
| CodingQwen3.5-122B-A10B wins | ||
| HumanEval | 82.6% | — |
| SWE-bench Pro | 55% | — |
| SWE-bench Verified | — | 72% |
| LiveCodeBench | — | 78.9% |
| Multimodal & GroundedQwen3.5-122B-A10B wins | ||
| MMMU-Pro | 54% | 76.9% |
| OfficeQA Pro | 38% | — |
| ReasoningQwen3.5-122B-A10B wins | ||
| LongBench v2 | 30% | 60.2% |
| MRCRv2 | 33% | — |
| KnowledgeQwen3.5-122B-A10B wins | ||
| MMLU | 84.8% | — |
| GPQA | 56.1% | 86.6% |
| FrontierScience | 52% | — |
| MMLU-Pro | — | 86.7% |
| SuperGPQA | — | 67.1% |
| Instruction Following | ||
| IFEval | — | 93.4% |
| MultilingualQwen3.5-122B-A10B wins | ||
| MGSM | 80.6% | — |
| MMLU-ProX | 60% | 82.2% |
| Mathematics | ||
| MATH-500 | 94.6% | — |
Qwen3.5-122B-A10B is ahead overall, 71 to 49. The biggest single separator in this matchup is GPQA, where the scores are 56.1% and 86.6%.
Qwen3.5-122B-A10B has the edge for knowledge tasks in this comparison, averaging 81.6 versus 53.6. Inside this category, GPQA is the benchmark that creates the most daylight between them.
Qwen3.5-122B-A10B has the edge for coding in this comparison, averaging 76.3 versus 55. Phi-4 stays close enough that the answer can still flip depending on your workload.
Qwen3.5-122B-A10B has the edge for reasoning in this comparison, averaging 60.2 versus 31.4. Inside this category, LongBench v2 is the benchmark that creates the most daylight between them.
Qwen3.5-122B-A10B has the edge for agentic tasks in this comparison, averaging 56 versus 38.3. Inside this category, BrowseComp is the benchmark that creates the most daylight between them.
Qwen3.5-122B-A10B has the edge for multimodal and grounded tasks in this comparison, averaging 76.9 versus 46.8. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.
Qwen3.5-122B-A10B has the edge for multilingual tasks in this comparison, averaging 82.2 versus 67.2. 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.