NLCO - Set-S - Feasibility Rate: leaderboard
Metric: Average feasibility rate: output satisfies every constraint (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O4 Mini (High) | 98.9 |
| 2 | GPT-5.1 (Medium) | 98.9 |
| 3 | Gemini 3 Flash (High) | 98.8 |
| 4 | Claude Sonnet 4.5 (Thinking) | 98.7 |
| 5 | DeepSeek V3.2 (Thinking) | 98.7 |
| 6 | Grok 4.1 Fast (Reasoning) | 98.3 |
| 7 | Qwen 3 235B A22B 2507 Instruct | 95.3 |
| 8 | DeepSeek V3.2 (Non-reasoning) | 93.4 |
| 9 | Qwen 3 14B (Reasoning) | 90.7 |
| 10 | QwQ-32B | 89.9 |
| 11 | Llama 4 Maverick Instruct | 78.8 |
| 12 | MiMo-V2-Flash (Non-reasoning) | 75 |
| 13 | Ministral-3-14B-Instruct-2512 | 72 |
| 14 | Qwen 3 14B (Non-reasoning) | 70.4 |
Interactive version: theaggregate.ai/benchmark?slug=nlco-set-s-feasibility-rate · How It Works · Data refreshed daily, snapshot 2026-09-25.