IFEval leaderboard
As of November 14, 2026, among the results listed here, LiquidAI LFM2.5-24B-A2B has the highest reported IFEval score, 96.8, published by Liquid, ahead of LiquidAI LFM3-1.7B at 96.7. 13 models have a IFEval result in our records.
Scores as the model makers published them, not re-run by us; each row links to its announcement.
13 models ranked · 8 makers
| Rank | Model | Score | Setting | Reported by | Date |
|---|---|---|---|---|---|
| 1 | LFM2.5-24B-A2B | 96.8 | Not published | Liquid | October 13, 2026 (model release) |
| 2 | LFM3-1.7B | 96.7 | Not published | Liquid | November 6, 2026 (model release) |
| 3 | LFM2.5-2.6BRetired | 95.4 | Not published | Liquid | November 6, 2026 (model release) |
| 4 | Qwen3.7 Plus | 94.6 | Not published | Qwen | May 21, 2026 |
| 5 | Qwen3.6 Plus | 94.3 | Not published | Qwen | May 21, 2026 |
| 6 | Llama 3.3 70B Instruct | 92.1 | Not published | Meta Llama | December 6, 2024 (model release) |
| 7 | P Ternary Bonsai 2 27B | 91.3 | Not published | Prism Ml | September 18, 2026 (model release) |
| 8 | LongCat 2.0 | 90 | Not published | Meituan | July 20, 2026 (model release) |
| 9 | Kimi K2 0711 | 89.8 | Instruction model evaluation results; Prompt Strict | Moonshot AI | July 11, 2025 (model release) |
| 10 | Llama 3.1 70B Instruct | 87.5 | Not published | Meta Llama | December 6, 2024 (model release) |
| 11 | GPT-4.1 | 87.4 | Not published | OpenAI | May 28, 2023 (model release) |
| 12 | Llama 3.1 8B Instruct | 80.4 | Not published | Meta Llama | December 6, 2024 (model release) |
| 13 | H SmolLM3-3B | 71.2 | Instruction Model; Extended Thinking | HuggingFaceTB | Publication date not reported |
Other reported runs
| Model | Score | Setting | Reported by | Date |
|---|---|---|---|---|
| H SmolLM3-3B | 76.7 | Instruction Model; No Extended Thinking | HuggingFaceTB | Publication date not reported |
| Llama 3.1 8B Instruct | 80.4 | Instruction Tuned Models; 0 | Meta Llama | Publication date not reported |
How much to trust these numbers
- Source
- Reported by the model makers themselves. Not re-run by LLMBenchmark or anyone independent.
- Settings
- Settings differ between makers. Check the Setting column before comparing two scores.
- Spread
- 96.8 to 71.2 across 13 models.
- Freshness
- Newest result published November 6, 2026; oldest May 28, 2023.
What IFEval measures
IFEval asks models to follow verifiable instructions, and its score indicates how well they follow those instructions. Source: Instruction-Following Evaluation for Large Language Models