Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Qwen2.5-72B
60
2/8 categoriesSarvam 105B
60
3/8 categoriesQwen2.5-72B· Sarvam 105B
Treat this as a split decision. Qwen2.5-72B makes more sense if agentic is the priority or you would rather avoid the extra latency and token burn of a reasoning model; Sarvam 105B is the better fit if knowledge is the priority or you want the stronger reasoning-first profile.
Qwen2.5-72B and Sarvam 105B finish on the same overall score, so this is less about a single winner and more about where the edge shows up. The headline says tie; the benchmark table is where the real choice happens.
Sarvam 105B is the reasoning model in the pair, while Qwen2.5-72B 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 | Qwen2.5-72B | Sarvam 105B |
|---|---|---|
| AgenticQwen2.5-72B wins | ||
| Terminal-Bench 2.0 | 56% | — |
| BrowseComp | 64% | 49.5% |
| OSWorld-Verified | 55% | — |
| CodingSarvam 105B wins | ||
| HumanEval | 75% | — |
| SWE-bench Verified | 46% | 45% |
| LiveCodeBench | 40% | — |
| SWE-bench Pro | 47% | — |
| LiveCodeBench v6 | — | 71.7% |
| Multimodal & Grounded | ||
| MMMU-Pro | 64% | — |
| OfficeQA Pro | 70% | — |
| Reasoning | ||
| MuSR | 78% | — |
| BBH | 81% | — |
| LongBench v2 | 72% | — |
| MRCRv2 | 71% | — |
| gpqaDiamond | — | 78.7% |
| hle | — | 11.2% |
| KnowledgeSarvam 105B wins | ||
| MMLU | 83% | 90.6% |
| GPQA | 82% | — |
| SuperGPQA | 80% | — |
| MMLU-Pro | 75% | 81.7% |
| HLE | 11% | — |
| FrontierScience | 70% | — |
| SimpleQA | 80% | — |
| Instruction FollowingQwen2.5-72B wins | ||
| IFEval | 85% | 84.8% |
| Multilingual | ||
| MGSM | 84% | — |
| MMLU-ProX | 79% | — |
| MathematicsSarvam 105B wins | ||
| AIME 2023 | 84% | — |
| AIME 2024 | 86% | — |
| AIME 2025 | 85% | 88.3% |
| HMMT Feb 2023 | 80% | — |
| HMMT Feb 2024 | 82% | — |
| HMMT Feb 2025 | 81% | — |
| BRUMO 2025 | 83% | — |
| MATH-500 | 84% | 98.6% |
| HMMT Feb 2025 | — | 85.8% |
| HMMT Nov 2025 | — | 85.8% |
Qwen2.5-72B and Sarvam 105B are tied on overall score, so the right pick depends on which category matters most for your use case.
Sarvam 105B has the edge for knowledge tasks in this comparison, averaging 81.7 versus 61.5. Inside this category, MMLU is the benchmark that creates the most daylight between them.
Sarvam 105B has the edge for coding in this comparison, averaging 45 versus 44.1. Inside this category, SWE-bench Verified is the benchmark that creates the most daylight between them.
Sarvam 105B has the edge for math in this comparison, averaging 92.3 versus 84.1. Inside this category, MATH-500 is the benchmark that creates the most daylight between them.
Qwen2.5-72B has the edge for agentic tasks in this comparison, averaging 57.7 versus 49.5. Inside this category, BrowseComp is the benchmark that creates the most daylight between them.
Qwen2.5-72B has the edge for instruction following in this comparison, averaging 85 versus 84.8. 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.