Is Thinking Longer Worth It?

Archived study

Examines the relationship between recorded AI evaluation scores and estimated token costs across thinking-effort settings. The archive includes score-versus-cost charts and an analysis script.

Interpretation and limits

Scores and estimated costs have different evidence bases. Cost parity is modeled rather than measured across a common live workload. Model identities, pricing, and evaluation claims in the original session are historical inputs, not independently verified current product facts.

Package and provenance

The original .science export and matching poster are preserved byte-for-byte from Hugging Face snapshot b0218fea4fc3fc9aa37ac616aba20e619e242c1e. The package contains the archived conversation, session metadata, and resources listed in its internal manifest. This introduction summarizes the archived record; the research, external sources, and model evaluations were not rerun for this reorganization.

Case files

• Session export
• Case poster