mapluisch/mlx-benchmark-leaderboard

Upload bench_openrouter_nvidia_nemotron-3-ultra-550b-a55b_20260612_143059.json

Upload 4 files

Upload 3 files

Delete data/bench_ollama_deepseek-v4-flash:cloud_20260427_115114.json

Upload bench_ollama_minimax-m3:cloud_20260602_140116.json

Upload bench_openrouter_qwen_qwen3-6-35b-a3b_20260428_100628.json

Delete data/bench_openrouter_qwen_qwen3-6-35b-a3b_20260428_100126.json

Delete data/bench_openrouter_qwen_qwen3-6-35b-a3b_20260428_100628.json

Upload 3 files

Upload bench_ollama_kimi-k2-6:cloud_20260427_174332.json

Rewrite app.py to parse real mlx-bench output format (stats.by_type/by_difficulty/by_category with total/correct counts). Updated metadata cols, submit tab with real format example."

Upload 15 files

Add real benchmark result: GPT-5 Nano (real mlx-bench output format)

Remove old placeholder data files (wrong format)

Add app.py - MLX Benchmark V2 Leaderboard with interactive table, charts, and submission info

Add data: Llama 3.3 70B results

Add data: Qwen3-8B results

Add data: Llama 3.2 3B results

Add data: JOSIE-4B-Thinking results

Add data: Gemini 2.5 Flash results

Add data: Claude Sonnet 4 results

Add data: GPT-4o results

Add data: Qwen3-32B results

Add requirements

Initial setup: README with Space metadata

initial commit