Token Usage and Cost for 9 LLMs on an 800-Household Test Case

800 household test cases were used to measure token consumption and associated costs for 9 different large language models. The dataset was authored by Michael Jones and published on figshare. Cost figures are correct as of September 2025, with the record last updated in June 2026.

Use Cases

• Compare cost per query across different LLMs based on the token usage data.
• Benchmark model efficiency for a specific application scenario based on the 800-household test case.
• Estimate budget requirements for deploying LLMs at scale using the provided cost-per-token metrics.

Strengths

• Costs are anchored to a specific point in time (September 2025), providing a historical benchmark.
• The test case is defined with a concrete scale of 800 households.
• The dataset compares 9 distinct LLMs, allowing for multi-model analysis.

Limitations

• The dataset is very small at 5.5 KB, indicating limited scope and detail.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

figshare

Collection Method

Likely generated from a benchmarking exercise or cost analysis.

Time Range

Costs are correct as of September 2025.

Freshness

Last updated 2026-06-02 17:34:02; freshness should be verified as costs are dynamic.