# LFM2.5-VL-3B vs LFM3-VL-2B

> Side-by-side benchmark comparison for LFM2.5-VL-3B and LFM3-VL-2B across agentic, coding, multimodal, knowledge, reasoning, multilingual, and math tasks.

- Canonical page: https://benchlm.ai/compare/lfm2-5-vl-3b-vs-lfm3-vl-2b
- Last updated: October 27, 2026

- Shared sourced benchmarks: 7
- HTML indexing: indexable
- Ranking lane: BenchAlign v5.8

## Quick Verdict

Pick LFM3-VL-2B if you want the stronger benchmark profile. LFM2.5-VL-3B only makes more sense when its price, context window, or workload-specific category wins matter more than the overall score.

## Summary

- LFM3-VL-2B leads overall 39 to 0.
- The biggest single benchmark swing is OCRBench V2 in Multimodal & Grounded, with scores of 47.5% and 63.5%.
- LFM3-VL-2B also has the larger context window at 128K.

## Model Snapshot

| Property | LFM2.5-VL-3B | LFM3-VL-2B |
|----------|----------|----------|
| Creator | LiquidAI | LiquidAI |
| Type | Open Weight | Open Weight |
| Reasoning | Non-Reasoning | Reasoning |
| Context | 32K | 128K |
| Overall Score | null | null |
| Benchmarks Covered | 10 | 20 |
| Pricing (input/output) | $0.00 / $0.00 | $0.00 / $0.00 |

## Category Breakdown

### Agentic

- Winner: Not comparable
- LFM2.5-VL-3B public-lane score: Coming soon
- LFM3-VL-2B public-lane score: Coming soon

| Benchmark | LFM2.5-VL-3B | LFM3-VL-2B | Winner |
|-----------|-----------|-----------|--------|
| BFCL v4 | 32.5% | Coming soon | Coming soon |

### Multimodal & Grounded

- Winner: Not comparable
- LFM2.5-VL-3B public-lane score: Coming soon
- LFM3-VL-2B public-lane score: 19.6 (Unranked · 19 rankable rows)

| Benchmark | LFM2.5-VL-3B | LFM3-VL-2B | Winner |
|-----------|-----------|-----------|--------|
| RealWorldQA | 73.1% | 79.3% | LFM3-VL-2B |
| SimpleVQA | 35.4% | 36.3% | LFM3-VL-2B |
| CountBench | 87.3% | 93.0% | LFM3-VL-2B |
| MMMU | 48.4% | 55.7% | LFM3-VL-2B |
| MMMU-Pro | 30.5% | 45.7% | LFM3-VL-2B |
| OCRBench V2 | 47.5% | 63.5% | LFM3-VL-2B |
| RefCOCO (avg) | 87.9% | 90.5% | LFM3-VL-2B |
| V* | Coming soon | 86.4% | Coming soon |
| CharXiv w/o tools | Coming soon | 70.2% | Coming soon |
| MathVision | Coming soon | 51.6% | Coming soon |
| DynaMath | Coming soon | 79.2% | Coming soon |
| We-Math | Coming soon | 68.6% | Coming soon |
| ERQA | Coming soon | 48.2% | Coming soon |
| ScreenSpot Pro | Coming soon | 55.5% | Coming soon |
| LVBench | Coming soon | 41.2% | Coming soon |
| MLVU (M-Avg) | Coming soon | 71.3% | Coming soon |
| MMVU | Coming soon | 52.9% | Coming soon |
| Video-MME (w/o subtitle) | Coming soon | 62.8% | Coming soon |
| Video-MME (with subtitle) | Coming soon | 69.0% | Coming soon |
| VideoMMMU | Coming soon | 49.6% | Coming soon |

### Instruction Following

- Winner: Not comparable
- LFM2.5-VL-3B public-lane score: Coming soon
- LFM3-VL-2B public-lane score: Coming soon

| Benchmark | LFM2.5-VL-3B | LFM3-VL-2B | Winner |
|-----------|-----------|-----------|--------|
| IFEval | 82.3% | Coming soon | Coming soon |
| IFBench | 25.8% | Coming soon | Coming soon |

## FAQ

### Which is better overall, LFM2.5-VL-3B or LFM3-VL-2B?

LFM3-VL-2B is ahead overall on BenchLM right now.

### How many benchmarks does LFM2.5-VL-3B cover on BenchLM?

LFM2.5-VL-3B currently has 10 sourced benchmark scores on BenchLM.

### How many benchmarks does LFM3-VL-2B cover on BenchLM?

LFM3-VL-2B currently has 20 sourced benchmark scores on BenchLM.

## Related Comparisons

- [LFM2.5-VL-3B vs Claude Opus 5.5](/compare/claude-opus-5-5-vs-lfm2-5-vl-3b)
- [LFM3-VL-2B vs Claude Opus 5.5](/compare/claude-opus-5-5-vs-lfm3-vl-2b)
- [LFM2.5-VL-3B vs GPT-6 Astra](/compare/gpt-6-astra-vs-lfm2-5-vl-3b)
- [LFM3-VL-2B vs GPT-6 Astra](/compare/gpt-6-astra-vs-lfm3-vl-2b)

## Explore More

- [LFM2.5-VL-3B profile](/models/lfm2-5-vl-3b)
- [LFM3-VL-2B profile](/models/lfm3-vl-2b)
- [Compare Pricing](/llm-pricing)
- [Alternative Finder](/tools/alternative-finder)
- [LLM Selector](/tools/llm-selector)
- [Overall Rankings](/best/overall)