Flow - 4x faster than typing
Mini toolsIFBench
Instruction Following Benchmark
Measures how reliably a model follows complex multi-constraint instructions, a known weak spot for many otherwise strong models.
Language
Text
Accuracy
Max 100.0%
Released Jun 2025
5
Results
5
Models scored
76.1%
Top: Qwen 3.5 122B A10B
70.2%
Median
Best results
Top primary scores; one row per model.
Frontier over time
Each dot is one model result; the line traces the running best score.
All results
Showing one canonical row per model. Show all configurations
| # | Model | Score | Conditions | Eval date | Source | Flags |
|---|---|---|---|---|---|---|
| 1 | Qwen 3.5 122B A10B | 76.1% | — | 24 Apr 2026 | Third-party | Primary Verified |
| 2 | Nemotron 3 | 71.5% | standard | 15 Dec 2025 | Self-reported | Primary |
| 3 | Qwen 3.5 35B A3B | 70.2% | — | 15 Feb 2025 | Third-party | Primary Verified |
| 4 | Kimi K2.7 Code | 63.1% | 0-shot · standard | 12 Jun 2026 | Third-party | Primary Verified |
| 5 | Trinity Large Thinking | 52.3% | 0-shot · standard | 01 Apr 2026 | Self-reported | Primary |
