Liquid LFM2.5-2.6B Card - IFStruct: leaderboard
IFStruct is Liquid AI's open benchmark of structured-output compliance: prompts ask for JSON or YAML items with schema requirements stated in varied ways, and a sample passes only if it parses and meets every schema check. LFM2.5-2.6B model card, every model run by Liquid AI with vLLM; percent of passing samples, higher is better.
Metric: Pass Rate (%). Source: huggingface.co. Saturation forecast: Around February 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LFM2.5-2.6B | 85.49 | #836 |
| 2 | Qwen 3.5 9B | 78.5 | #391 |
| 3 | gemma-4-E4B-it | 76.65 | #577 |
| 4 | Gemma 4 E2B (IT) | 64.85 | #779 |
| 5 | Qwen 3.5 4B | 36.25 | #514 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=liquid-lfm2-5-2-6b-card-ifstruct · How It Works · Data refreshed daily, snapshot 2026-11-05.