Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Gemini 3 Pro
79
Winner · 4/8 categoriesSarvam 105B
60
0/8 categoriesGemini 3 Pro· Sarvam 105B
Pick Gemini 3 Pro if you want the stronger benchmark profile. Sarvam 105B only becomes the better choice if you want the stronger reasoning-first profile.
Gemini 3 Pro is clearly ahead on the aggregate, 79 to 60. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Gemini 3 Pro's sharpest advantage is in agentic, where it averages 71.1 against 49.5. The single biggest benchmark swing on the page is BrowseComp, 83% to 49.5%.
Sarvam 105B is the reasoning model in the pair, while Gemini 3 Pro 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. Gemini 3 Pro gives you the larger context window at 2M, compared with 128K for Sarvam 105B.
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 | Gemini 3 Pro | Sarvam 105B |
|---|---|---|
| AgenticGemini 3 Pro wins | ||
| Terminal-Bench 2.0 | 68% | — |
| BrowseComp | 83% | 49.5% |
| OSWorld-Verified | 66% | — |
| CodingGemini 3 Pro wins | ||
| HumanEval | 91% | — |
| SWE-bench Verified | 59% | 45% |
| LiveCodeBench | 49% | — |
| SWE-bench Pro | 58% | — |
| LiveCodeBench v6 | — | 71.7% |
| Multimodal & Grounded | ||
| MMMU | 87.2% | — |
| MMMU-Pro | 81% | — |
| OfficeQA Pro | 92% | — |
| RealWorldQA | 83.3% | — |
| OmniDocBench 1.5 | 88.5% | — |
| MathVision | 86.6% | — |
| We-Math | 86.9% | — |
| DynaMath | 85.1% | — |
| MStar | 83.1% | — |
| SimpleVQA | 73.2% | — |
| ChatCVQA | 81.4% | — |
| MMLongBench-Doc | 60.5% | — |
| CC-OCR | 79.0% | — |
| AI2D_TEST | 94.1% | — |
| CountBench | 97.3% | — |
| RefCOCO (avg) | 84.1% | — |
| ODINW13 | 46.3% | — |
| ERQA | 70.5% | — |
| VideoMMMU | 87.6% | — |
| MLVU (M-Avg) | 83.0% | — |
| ScreenSpot Pro | 72.7% | — |
| Reasoning | ||
| MuSR | 93% | — |
| BBH | 90% | — |
| LongBench v2 | 90% | — |
| MRCRv2 | 87% | — |
| ARC-AGI-2 | 31.1% | — |
| gpqaDiamond | — | 78.7% |
| hle | — | 11.2% |
| KnowledgeGemini 3 Pro wins | ||
| MMLU | 99% | 90.6% |
| GPQA | 97% | — |
| SuperGPQA | 95% | — |
| FrontierScience | 86% | — |
| SimpleQA | 95% | — |
| MMLU-Pro | — | 81.7% |
| Instruction Following | ||
| IFEval | — | 84.8% |
| Multilingual | ||
| Coming soon | ||
| MathematicsGemini 3 Pro wins | ||
| AIME 2023 | 99% | — |
| AIME 2024 | 99% | — |
| AIME 2025 | 98% | 88.3% |
| HMMT Feb 2023 | 95% | — |
| HMMT Feb 2024 | 97% | — |
| HMMT Feb 2025 | 96% | — |
| BRUMO 2025 | 96% | — |
| MATH-500 | — | 98.6% |
| HMMT Feb 2025 | — | 85.8% |
| HMMT Nov 2025 | — | 85.8% |
Gemini 3 Pro is ahead overall, 79 to 60. The biggest single separator in this matchup is BrowseComp, where the scores are 83% and 49.5%.
Gemini 3 Pro has the edge for knowledge tasks in this comparison, averaging 92.5 versus 81.7. Inside this category, MMLU is the benchmark that creates the most daylight between them.
Gemini 3 Pro has the edge for coding in this comparison, averaging 54.8 versus 45. Inside this category, SWE-bench Verified is the benchmark that creates the most daylight between them.
Gemini 3 Pro has the edge for math in this comparison, averaging 97.1 versus 92.3. Inside this category, AIME 2025 is the benchmark that creates the most daylight between them.
Gemini 3 Pro has the edge for agentic tasks in this comparison, averaging 71.1 versus 49.5. Inside this category, BrowseComp 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.