TuRTLe Spec-to-RTL (Verilator): leaderboard
TuRTLe leaderboard variant for spec-to-RTL generation evaluated with Verilator across VerilogEval S2R and RTLLM.
Metric: Average Score (%). Source: huggingface.co. Status: saturation imminent. 45 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K2.5 | 78.8 |
| 2 | DeepSeek R1 0528 | 76.69 |
| 3 | GLM-5 FP8 | 75.67 |
| 4 | DeepSeek R1 | 74.74 |
| 5 | Gemma 4 31B (IT) | 74.15 |
| 6 | DeepSeek V3.1 Terminus | 73.59 |
| 7 | GPT-OSS-120B | 71.28 |
| 8 | GPT-OSS-20B | 62.49 |
| 9 | QwQ-32B | 57.97 |
| 10 | Llama 3.1 405B | 56.17 |
| 11 | Qwen 2.5 72B | 52.21 |
| 12 | Qwen 2.5 32B | 50.13 |
| 13 | starchat2-15B-v0.1 | 46.86 |
| 14 | Qwen 3 8B | 45.08 |
| 15 | Hermes-4-14B | 42.61 |
Interactive version: theaggregate.ai/benchmark?slug=turtle-spec-to-rtl-verilator · How It Works · Data refreshed daily, snapshot 2026-09-05.