BALSAM - Text Manipulation: 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) | 74.07 |
| 2 | GLM-5 (Thinking) | 65.08 |
| 3 | Command-R+ | 64.69 |
| 4 | Command A 03 2025 | 64.34 |
| 5 | GPT-5.2 | 64.32 |
| 6 | Grok 4.1 Fast (Reasoning) | 64.12 |
| 7 | Kimi K2 Instruct (0905) | 63.73 |
| 8 | AceGPT-v2-8B-Chat | 62.26 |
| 9 | DeepSeek V3.2 | 61.92 |
| 10 | DeepSeek V4 Pro (Reasoning) | 61.82 |
| 11 | Fanar-C-2-27B | 60.57 |
| 12 | GLM-4.7 (Reasoning) | 60.19 |
| 13 | Llama 3.3 70B Instruct | 59.49 |
| 14 | Qwen 3 235B A22B | 59.32 |
| 15 | Llama 4 Maverick Instruct | 57.96 |
Interactive version: theaggregate.ai/benchmark?slug=balsam-text-manipulation · How It Works · Data refreshed daily, snapshot 2026-09-19.