2026-10-12 · Mac · Apple M5 Max
A focused rerun on M5 Max · Apple M5 Max (integrated GPU) · 128 GB: 2 models across 4 task sets. Every round is listed in the model scorecard, with its results side by side in one table.
M5 Max · Apple M5 Max (integrated GPU) · 128 GB · anthropic-cli + codex-cli · focused subset · 1 trial per task · 2026-10-12 · gezel 5b41f675e65b · catalog 0.1.94
Tests Claude CLI and Codex CLI again on the nine tasks where they ran into trouble on October 11, plus two they had passed, after fixes to gezel and to our tests. Each task ran once on each app. The Research to Word task was allowed 45 minutes instead of 30. This covers 11 tasks, not the full set.
General capability
| Model | Size | Tasks passed | Performance |
|---|---|---|---|
| Claude Sonnet 5.5 · Claude CLI | cloud | 4/4 (some tasks run once — count not rate) | — |
| GPT-6 Luna · Codex CLI | cloud | 4/4 (some tasks run once — count not rate) | — |
Tasks in this set: 4.
Office and knowledge work
| Model | Size | Tasks passed | Performance |
|---|---|---|---|
| Claude Sonnet 5.5 · Claude CLI | cloud | 4/4 (some tasks run once — count not rate) | — |
| GPT-6 Luna · Codex CLI | cloud | 4/4 (some tasks run once — count not rate) | — |
Tasks in this set: 4.
Engineering work
| Model | Size | Tasks passed | Performance |
|---|---|---|---|
| Claude Sonnet 5.5 · Claude CLI | cloud | 1/1 (some tasks run once — count not rate) | — |
| GPT-6 Luna · Codex CLI | cloud | 1/1 (some tasks run once — count not rate) | — |
Tasks in this set: 1.
Complex workflows
| Model | Size | Tasks passed | Performance |
|---|---|---|---|
| Claude Sonnet 5.5 · Claude CLI | cloud | 2/2 (some tasks run once — count not rate) | — |
| GPT-6 Luna · Codex CLI | cloud | 2/2 (some tasks run once — count not rate) | — |
Tasks in this set: 2.