Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
1-bit Bonsai 4B
~44
1/8 categoriesQwen3 235B 2507
48
Winner · 3/8 categories1-bit Bonsai 4B· Qwen3 235B 2507
Pick Qwen3 235B 2507 if you want the stronger benchmark profile. 1-bit Bonsai 4B only becomes the better choice if mathematics is the priority.
Qwen3 235B 2507 is clearly ahead on the aggregate, 48 to 44. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Qwen3 235B 2507's sharpest advantage is in knowledge, where it averages 63.8 against 28.7. The single biggest benchmark swing on the page is GPQA, 28.7% to 77.5%. 1-bit Bonsai 4B does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.
Qwen3 235B 2507 gives you the larger context window at 128K, compared with 32K for 1-bit Bonsai 4B.
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 | 1-bit Bonsai 4B | Qwen3 235B 2507 |
|---|---|---|
| Agentic | ||
| Terminal-Bench 2.0 | — | 33% |
| BrowseComp | — | 40% |
| OSWorld-Verified | — | 30% |
| Coding | ||
| HumanEval | — | 31% |
| SWE-bench Verified | — | 15% |
| LiveCodeBench | — | 51.8% |
| SWE-bench Pro | — | 19% |
| Multimodal & Grounded | ||
| MMMU-Pro | — | 38% |
| OfficeQA Pro | — | 46% |
| ReasoningQwen3 235B 2507 wins | ||
| MuSR | 41.4% | 35% |
| BBH | — | 60% |
| LongBench v2 | — | 52% |
| MRCRv2 | — | 52% |
| KnowledgeQwen3 235B 2507 wins | ||
| GPQA | 28.7% | 77.5% |
| MMLU | — | 39% |
| SuperGPQA | — | 62.6% |
| MMLU-Pro | — | 83% |
| FrontierScience | — | 39% |
| SimpleQA | — | 54.3% |
| Instruction FollowingQwen3 235B 2507 wins | ||
| IFEval | 69.6% | 88.7% |
| Multilingual | ||
| MGSM | — | 63% |
| MMLU-ProX | — | 79.4% |
| Mathematics1-bit Bonsai 4B wins | ||
| MATH-500 | 65.8% | 57% |
| AIME 2023 | — | 39% |
| AIME 2024 | — | 41% |
| AIME 2025 | — | 70.3% |
| HMMT Feb 2023 | — | 35% |
| HMMT Feb 2024 | — | 37% |
| HMMT Feb 2025 | — | 36% |
| BRUMO 2025 | — | 38% |
Qwen3 235B 2507 is ahead overall, 48 to 44. The biggest single separator in this matchup is GPQA, where the scores are 28.7% and 77.5%.
Qwen3 235B 2507 has the edge for knowledge tasks in this comparison, averaging 63.8 versus 28.7. Inside this category, GPQA is the benchmark that creates the most daylight between them.
1-bit Bonsai 4B has the edge for math in this comparison, averaging 65.8 versus 55.7. Inside this category, MATH-500 is the benchmark that creates the most daylight between them.
Qwen3 235B 2507 has the edge for reasoning in this comparison, averaging 47.5 versus 41.4. Inside this category, MuSR is the benchmark that creates the most daylight between them.
Qwen3 235B 2507 has the edge for instruction following in this comparison, averaging 88.7 versus 69.6. Inside this category, IFEval 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.