TokenSave Agent Token Measurements

Project Oct 18, 2026

TokenSave's published two-run comparison of Claude Code and Codex startup overhead, MCP tools and a browser-game coding task.

Visit primary link 2 sources attached

TokenSave's chart compares Claude Code and Codex startup, MCP startup and browser-game task token counts from two runs per setting. — TokenSave's October 18 two-run measurements on one Windows laptop. Different default models and cached input are included; this is not a general cost or quality ranking. Full author chart preserved, not independently reproduced by AITrending. Source: TokenSave (CC BY 4.0)

coding-agents agent-ecosystem verification

What happened lately

TokenSave Agent Token Measurements timeline

Ecosystem Oct 18, 2026

TokenSave says Codex used half the input tokens in a tiny Claude Code showdown

Community reports 3 sources

TokenSave published this CC BY 4.0 dataset on October 18, 2026. It provides prompts, per-run token counts and setup details for Claude Code 2.1.295 with Opus 5.5 and Codex CLI 0.162 with gpt-6-luna on one Windows laptop. The author also operates a usage-tracking product.

Its two runs per setting and single coding task support a narrow comparison of those configurations. Cached tokens, different models and installed plugins prevent treating the results as a general cost or coding-quality ranking. AITrending reviewed the published data without rerunning the experiment.