MLX Benchmark V2 - Medium — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 85.8 |
| 2 | Claude Sonnet 4.6 | 85.08 |
| 3 | Gemma 4 26B A4B (IT) | 82.32 |
| 4 | Grok 4.1 Fast | 78.45 |
| 5 | DeepSeek V4 Pro | 77.09 |
| 6 | Qwen 3.6 Max Preview | 76.67 |
| 7 | GPT-5.4 Nano | 75.14 |
| 8 | Gemini 2.5 Flash Lite (Preview 09-2025) | 71.82 |
| 9 | DeepSeek V4 Flash | 71.27 |
| 10 | GPT-5 Nano | 49.17 |
| 11 | Qwen 3.6 35B A3B | 46.41 |
| 12 | GLM-5.1 | 18.78 |
| 13 | Nemotron 3 Ultra | 17.13 |
| 14 | Kimi K2.5 | 4.97 |
| 15 | Kimi K2.6 | 2.76 |
Interactive version: theaggregate.ai/benchmark?slug=mlx-benchmark-v2-medium · How the rankings work · Data refreshed daily, snapshot 2026-07-22.