MLX Benchmarks vs Super PiBench

MLX Benchmark helps you understand how large language models perform on your hardware. Select a model and quantization, set context length and iterations, and run repeatable benchmarks that capture load time, tokenization time, time to first token, throughput, memory usage, and jitter. View detailed metrics grids, visualize comparisons in charts, and manage downloaded model repos directly in the app. Built for developers and AI enthusiasts who want clear, on-device performance insights without leaving their device. Features: - Model selection with quantization details - Configurable prompt, context length, and iterations - Metrics for load, tokenization, first-token latency, throughput, jitter, and memory - Charts to compare runs and devices - Downloaded model management (list, delete, refresh) - Optional Supabase upload for aggregating benchmark results

Super PiBench computes Pi to any number of decimal places using the Chudnovsky algorithm — one of the fastest known methods. Built on a high-performance C/GMP engine for dramatically fast computation, comprehensive benchmarking, and a global leaderboard to see how your device stacks up worldwide. GLOBAL LEADERBOARD (Pro) - All-time board + weekly board: refreshed every Monday, newcomers can reach the top too - Submit your benchmark score to compete with users worldwide - See where your Apple device ranks among iPhones, iPads, Macs, Apple Watches, and more - Scores verified on-device and uploaded securely via iCloud HIGH-PERFORMANCE ENGINE - Built on GMP (GNU Multiple Precision Arithmetic Library) for optimized large-number arithmetic - Chudnovsky algorithm with binary splitting: ~14 digits per term - Multi-core parallel computation with threaded binary splitting and merge (Pro) - More accurate time estimation with per-magnitude calibration APPLE WATCH - Compute Pi right on your wrist — works without iPhone - Multi-core computation and global leaderboard submission (Pro) - Digit limit raised to 10 million (Pro) - One-tap submission after calculating, with built-in time estimation PRESETS - Free: 100 / 10,000 / 100,000 / 1,000,000 digits - Pro: 10,000,000 / 20,000,000 / 50,000,000 / 100,000,000 digits - Custom input for any digit count STRESS TEST - Push your chip to the limit with repeated benchmark rounds - Free: up to 10 million digits, 5 rounds, single-core - Pro: up to 100 million digits, 20 rounds, multi-core - Bar chart visualization with color-coded best/worst results - Automatic trend detection: stable, throttling, or improving TOOLS & SHARING - Search for any number sequence with fast Boyer-Moore algorithm (Pro) - Generate and share beautiful computation result cards (Pro) - Save computation history for later review (Pro) - Copy, share, or export results with one tap KNOWLEDGE - Pi Knowledge module: fun facts, historical timeline, real-world applications, and algorithm explanations LANGUAGES - 14 languages: English, Simplified Chinese, Traditional Chinese, Japanese, Korean, German, French, Spanish, Portuguese, Italian, Dutch, Russian, Arabic, Greek DESIGN - iOS 26 Liquid Glass design: floating tab bar and glass effects throughout - Dark and light mode support - Core functionality runs entirely on-device, no internet required - Universal app: iPhone, iPad, Mac (Catalyst), and Apple Watch Free version computes up to 1,000,000 digits. Unlock Pro for global leaderboard, multi-core computing, extended stress test, digit search, history, share cards, and more.