MLX Benchmark V2 - MLX Embeddings LoRA: leaderboard

MLX Benchmark V2 category score for MLX embedding LoRA workflows.

Metric: Accuracy (%). Source: huggingface.co. 22 models tracked.

Top models

#ModelScoreOverall rank
1GPT-5.4 Nano86.67#362
2Claude Sonnet 4.666.67#96
3Gemini 3 Flash (Preview)60#90
4Gemma 4 26B A4B (IT)60#298
5Gemini 2.5 Flash Lite (Preview 09-2025)60#327
6DeepSeek V4 Flash53.33#155
7Grok 4.1 Fast53.33#227
8GPT-5 Nano46.67#438
9Qwen 3.6 35B A3B33.33#239
10Qwen 3.6 Max Preview33.33#93
11DeepSeek V4 Pro30.77#112
12Kimi K2.56.67#153
13Kimi K2.66.67#113
14GLM-5.16.67#109
15Nemotron 3 Ultra6.67#249

Interactive version: theaggregate.ai/benchmark?slug=mlx-benchmark-v2-mlx-embeddings-lora · How It Works · Data refreshed daily, snapshot 2026-11-06.