Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Claude 3 Haiku
42
Winner · 1/8 categoriesLFM2.5-350M
~39
1/8 categoriesClaude 3 Haiku· LFM2.5-350M
Pick Claude 3 Haiku if you want the stronger benchmark profile. LFM2.5-350M only becomes the better choice if instruction following is the priority.
Claude 3 Haiku has the cleaner overall profile here, landing at 42 versus 39. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
Claude 3 Haiku's sharpest advantage is in knowledge, where it averages 43.5 against 23.8. The single biggest benchmark swing on the page is MMLU-Pro, 63% to 20.0%. LFM2.5-350M does hit back in instruction following, so the answer changes if that is the part of the workload you care about most.
Claude 3 Haiku gives you the larger context window at 200K, compared with 32K for LFM2.5-350M.
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 | Claude 3 Haiku | LFM2.5-350M |
|---|---|---|
| Agentic | ||
| Terminal-Bench 2.0 | 40% | — |
| BrowseComp | 53% | — |
| OSWorld-Verified | 42% | — |
| Coding | ||
| HumanEval | 73% | — |
| SWE-bench Verified | 17% | — |
| LiveCodeBench | 20% | — |
| SWE-bench Pro | 19% | — |
| Multimodal & Grounded | ||
| MMMU-Pro | 70% | — |
| OfficeQA Pro | 67% | — |
| Reasoning | ||
| MuSR | 52% | — |
| BBH | 74% | — |
| LongBench v2 | 63% | — |
| MRCRv2 | 63% | — |
| KnowledgeClaude 3 Haiku wins | ||
| MMLU | 75.2% | — |
| GPQA | 56% | 30.6% |
| SuperGPQA | 54% | — |
| MMLU-Pro | 63% | 20.0% |
| HLE | 2% | — |
| FrontierScience | 50% | — |
| SimpleQA | 54% | — |
| Instruction FollowingLFM2.5-350M wins | ||
| IFEval | 76% | 77.0% |
| Multilingual | ||
| MGSM | 73% | — |
| MMLU-ProX | 70% | — |
| Mathematics | ||
| AIME 2023 | 56% | — |
| AIME 2024 | 58% | — |
| AIME 2025 | 57% | — |
| HMMT Feb 2023 | 52% | — |
| HMMT Feb 2024 | 54% | — |
| HMMT Feb 2025 | 53% | — |
| BRUMO 2025 | 55% | — |
| MATH-500 | 71% | — |
Claude 3 Haiku is ahead overall, 42 to 39. The biggest single separator in this matchup is MMLU-Pro, where the scores are 63% and 20.0%.
Claude 3 Haiku has the edge for knowledge tasks in this comparison, averaging 43.5 versus 23.8. Inside this category, MMLU-Pro is the benchmark that creates the most daylight between them.
LFM2.5-350M has the edge for instruction following in this comparison, averaging 77 versus 76. 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.