AI Models on
Apple M5 Max
20+ open-source models benchmarked on real hardware. Every number measured, not estimated. Interactive tools to explore the data.
Read the Benchmarks
Same data, five different perspectives. Pick the voice that matches how you like to read.
Benchmark Explorer
The full-featured version. Interactive comparison tools, RAM calculator, memory grid, sortable tables, cost analysis, 14-item FAQ.
Open benchmark explorer → CasualBlog Post Style
First-person, conversational. Written like telling a friend about a weekend project. Honest, including the parts that didn't work.
Read the blog version → DeveloperTechnical Reference
Terse, code-heavy. Assumes you know what quantization and MoE mean. Includes install commands, HuggingFace model IDs, and the benchmark formula.
Read the dev version → EditorialOpinion Column
Strong takes backed by data. Challenges conventional wisdom, calls out what's overrated, and tells you exactly what to buy (and what to skip).
Read the editorial → ProfessionalEnterprise Report
Formal analysis with executive summary. Frames results as ROI, TCO, and data sovereignty. The version you'd send to a CTO.
Read the report →Explore the Data
Visual dashboards and analysis tools built on the same benchmark dataset.
Interactive Dashboard
Visual benchmark dashboard with performance charts, cost comparison calculator, RAM tier guide, and hardware specs overview.
Open dashboard → AnalyzerModel Analyzer
Side-by-side model comparison tool, RAM calculator with visual memory map, speed charts, and filterable model grid.
Open analyzer → NEWOpen Model Leaderboard
Quality vs speed scatter plots, quantization explorer, local vs cloud comparison, and hardware tier recommendations. Inspired by Artificial Analysis.
Open leaderboard → LIVEData Dashboard
Live eval and benchmark queue status. Coverage matrix, model registry, run plans, and all raw data tables with filtering and sorting.
Open dashboard →