MLX Benchmark V2 - Multiple Choice — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 83.33 |
| 2 | Gemma 4 26B A4B (IT) | 83.33 |
| 3 | Kimi K2.5 | 83.33 |
| 4 | Gemini 3 Flash (Preview) | 83.33 |
| 5 | Qwen 3.6 35B A3B | 83.33 |
| 6 | Grok 4.1 Fast | 83.33 |
| 7 | Qwen 3.6 Max Preview | 83.33 |
| 8 | GLM-5.1 | 75 |
| 9 | DeepSeek V4 Pro | 75 |
| 10 | DeepSeek V4 Flash | 75 |
| 11 | Nemotron 3 Ultra | 75 |
| 12 | Gemini 2.5 Flash Lite (Preview 09-2025) | 66.67 |
| 13 | GPT-5.4 Nano | 58.33 |
| 14 | GPT-5 Nano | 50 |
| 15 | Kimi K2.6 | 50 |
Interactive version: theaggregate.ai/benchmark?slug=mlx-benchmark-v2-multiple-choice · How the rankings work · Data refreshed daily, snapshot 2026-07-22.