IFEval-TR - Turkish (strict): leaderboard
Measures verifiable instruction following in Turkish on IFEval-TR, a Turkish adaptation of Google's IFEval covering 910 examples and 25 constraint types across length, keyword and letter frequency, formatting, casing, punctuation and start/end markers. Every constraint is checked by a deterministic verifier; no LLM judge is involved. This series is the Turkish-specific set: 412 examples written from scratch around 50 Turkish cultural and historical topics. Strict scoring requires exact string matching with no morphological relaxation, which penalises morphologically equivalent Turkish responses. Scores are prompt-level accuracy. This series reports IFEval-TR - Turkish (strict) using Accuracy (%).
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around February 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B FP8 | 80 | #171 |
| 2 | Qwen 3.8 27B | 78.16 | #123 |
| 3 | Qwen 3.5 9B | 76.6 | #391 |
| 4 | GPT-OSS-120B | 63.26 | #346 |
| 5 | Gemma 3 27B (IT) | 61.41 | #528 |
| 6 | Gemma 3 12B (IT) | 59.95 | #672 |
| 7 | GPT-OSS-20B | 54.37 | #523 |
| 8 | Gemma 3 4B (IT) | 52.91 | #997 |
| 9 | GPT-4o Mini | 50 | #598 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ifeval-tr-turkish-strict · How It Works · Data refreshed daily, snapshot 2026-11-05.