MLX Benchmark V2 - Hard — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 92.66 |
| 2 | Gemini 3 Flash (Preview) | 77.78 |
| 3 | Qwen 3.6 Max Preview | 75 |
| 4 | GPT-5.4 Nano | 72.48 |
| 5 | DeepSeek V4 Flash | 72.48 |
| 6 | Gemma 4 26B A4B (IT) | 68.81 |
| 7 | DeepSeek V4 Pro | 65.56 |
| 8 | Grok 4.1 Fast | 65.14 |
| 9 | Gemini 2.5 Flash Lite (Preview 09-2025) | 64.22 |
| 10 | Qwen 3.6 35B A3B | 52.29 |
| 11 | GPT-5 Nano | 41.28 |
| 12 | Nemotron 3 Ultra | 12.84 |
| 13 | GLM-5.1 | 11.01 |
| 14 | Kimi K2.5 | 5.5 |
| 15 | Kimi K2.6 | 4.63 |
Interactive version: theaggregate.ai/benchmark?slug=mlx-benchmark-v2-hard · How the rankings work · Data refreshed daily, snapshot 2026-07-22.