# Multi-IF

> Meta benchmark of instruction following over multi-turn conversations in eight languages.

Canonical page: https://benchlm.ai/benchmarks/multiif

- Category: [Instruction Following](/instruction-following)
- Last updated: October 29, 2026

## About Multi-IF

- Year: 2024
- Tasks: Multi-turn, multilingual instructions
- Format: Instruction accuracy
- Difficulty: Keeping constraints across turns and languages
- Paper: [Multi-IF: Benchmarking LLMs on Multi-Turn and Multilingual Instructions Following](https://arxiv.org/abs/2410.15553)

Liquid AI reports it for LFM3-1.7B (October 27, 2026) at temperature 0.6 and top_p 0.95.

Multi-IF is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (1 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [LFM3-1.7B](/models/lfm3-1-7b) | LiquidAI | 89.10% |

## FAQ

### What does Multi-IF measure?

Meta benchmark of instruction following over multi-turn conversations in eight languages.

### Which model scores highest on Multi-IF?

LFM3-1.7B by LiquidAI currently leads with a score of 89.10% on Multi-IF.

### How many models are evaluated on Multi-IF?

1 AI models have published results on Multi-IF in the BenchLM catalog.

### Does Multi-IF affect BenchLM's overall score?

Not directly. Multi-IF is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.