BALSAM - Logic: leaderboard
Metric: Overall score (0-100, LLM-judged generation and multiple choice). Source: benchmarks.ksaa.gov.sa. 29 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 4 31B (IT) | 53.1 |
| 2 | GLM-5 (Thinking) | 42.41 |
| 3 | GLM-4.7 (Reasoning) | 40.51 |
| 4 | GPT-5.2 | 40.2 |
| 5 | MiniMax M3 (Reasoning) | 38.55 |
| 6 | Kimi K2 Instruct (0905) | 38.31 |
| 7 | AceGPT-v2-8B-Chat | 38.08 |
| 8 | Command A 03 2025 | 36.34 |
| 9 | Llama 4 Maverick Instruct | 36.33 |
| 10 | DeepSeek V3.2 | 35.75 |
| 11 | Qwen 3 235B A22B | 35.63 |
| 12 | Fanar-C-2-27B | 35.18 |
| 13 | Llama 3.3 70B Instruct | 35.06 |
| 14 | Grok 4.1 Fast (Reasoning) | 34.74 |
| 15 | Jais-2-70B-Chat | 34.01 |
Interactive version: theaggregate.ai/benchmark?slug=balsam-logic · How It Works · Data refreshed daily, snapshot 2026-09-19.