Creative diversity analysis: benchmark methods and results
Models analyse the same fixed account's reviewed creative labels and identify concentrations, gaps and useful next tests. The benchmark compares the analysis layer; it does not combine tagging accuracy with the headline analysis score.
Reviewed 16 October 2026 · Tests and editorial direction: Ben Nancarrow · AI-assisted writing and checks.
The account pack contains 35 files, including one exact duplicate, representing 34 unique assets. Both analysis conditions receive the fixed reviewed taxonomy records. One account is one case even when the model inspects many asset rows.
What changes between prompts
The simple and scaffolded analysis prompts use the same labelled portfolio. The scaffold structures the concentration, risk, gap and recommendation assessment. The earlier classification/instrument workflow is separate from this comparison.
How the result is scored
The evaluation checks analysis against an answer key, including numerical consistency, real risks, unsupported risks, gaps and deliberately absent categories. Recommendation quality and operator readiness have a separate judged layer. Read the published score as performance on this fixed analysis task, rather than proof that an arbitrary portfolio can be classified and planned end to end.
All 16 reviewed model results
Reviewed diversity-analysis task score / 10
Model
Result / setup
Scored outputs
Configuration / recorded test date
Cost / scored output
Claude Fable 5provisional
10.00/10Simple prompt
1
thinking_max29 August 2026
$12.939Mixed cost basis
Claude Fable 5.1provisional
10.00/10Simple prompt
1
thinking_max9 September 2026
$7.215Mixed cost basis
Claude Opus 4.8provisional
10.00/10Simple prompt
1
thinking_max29 August 2026
$3.461Mixed cost basis
Claude Opus 5provisional
10.00/10Simple prompt
1
thinking_max29 August 2026
$3.534Mixed cost basis
Claude Opus 5.5provisional
10.00/10Simple prompt
1
thinking_max23 September 2026
$14.539Mixed cost basis
Gemini 3.1 Pro Previewprovisional
10.00/10Scaffolded prompt
1
reasoning_high29 August 2026
$1.304Mixed cost basis
Gemini 3.5 Flashprovisional
10.00/10Scaffolded prompt
1
reasoning_high29 August 2026
$1.283Mixed cost basis
Gemini 3.7 Flashprovisional
10.00/10Simple prompt
1
reasoning_high29 August 2026
$0.542Mixed cost basis
Gemini 3.8 Flashprovisional
10.00/10Scaffolded prompt
1
reasoning_high9 September 2026
$0.507Recorded spend
Google Gemini 4 Argonprovisional
10.00/10Scaffolded prompt
1
high13 October 2026
$1.928Recorded spend
GPT-5.5provisional
10.00/10Simple prompt
1
reasoning_medium29 August 2026
$2.364Mixed cost basis
GPT-5.6 Solprovisional
10.00/10Simple prompt
1
xhigh29 August 2026
$4.053Mixed cost basis
GPT-6 Astraprovisional
10.00/10Scaffolded prompt
1
xhigh9 September 2026
$3.129Recorded spend
GPT-6 Solprovisional
10.00/10Simple prompt
1
xhigh23 September 2026
$1.674Recorded spend
Meta Muse Spark 1.3provisional
10.00/10Simple prompt
1
high8 September 2026
$0.557Recorded spend
Meta Muse Wonderprovisional
10.00/10Scaffolded prompt
1
max13 October 2026
$4.089Recorded spend
Recorded output counts can include repeats and both prompt conditions. Configurations and cost bases are retained; joint scores remain joint.
This record preserves the evidence and interpretation reviewed together. The live explorer can advance independently. Public results include provisional evidence; the methods above define what the scores measure.