Cadillac vs Audi
Which brand 4 AI models prefer, overall and by buyer need when available
Configurations compared: 2026 Cadillac CT5 (Auto (S10), 4 cyl, 2.0 L, Turbo) vs 2026 Audi A6 Allroad quattro (Auto (AM-S7), 6 cyl, 3.0 L, Turbo)
Simulation: scenario paths made up to reproduce the published per-judge shares, not measured rows.
Too close to call: the confidence interval includes 50%
Published 95% confidence interval for Cadillac: 41%–65%. Press Drop to watch simulated scenarios land.
How to read it
Each ball is a simulated buying scenario. Each row of pins is an AI model, pushing the ball toward the brand it picked.
One drop is one sample. Drop again and the pile shifts; the orange band shows where the balance lands in 95% of drops.
A leader is named only when the band clears the middle. The paths are simulated from the published per-judge shares: no model saw these scenarios.
- Other needs
Tap to drop. Hold to keep dropping.
How to read it
Each ball is a simulated buying scenario. Each row of pins is an AI model, pushing the ball toward the brand it picked.
One drop is one sample. Drop again and the pile shifts; the orange band shows where the balance lands in 95% of drops.
A leader is named only when the band clears the middle. The paths are simulated from the published per-judge shares: no model saw these scenarios.
The numbers behind the machine
AI consensus
Too close to call: the confidence interval includes 50%
Configurations compared: 2026 Cadillac CT5 (Auto (S10), 4 cyl, 2.0 L, Turbo) vs 2026 Audi A6 Allroad quattro (Auto (AM-S7), 6 cyl, 3.0 L, Turbo)
- Confidence interval for Cadillac: 41%–65%
- 95% confidence intervals (MOVER combination of the per-judge intervals, widened for small samples in battles and newer boards)
Orange: the 95% confidence interval for Cadillac. Notch: the point estimate. Line: 50%.
How the judges lean
- 1 judge leans to Cadillac
- 0 judges lean to Audi
- 3 judges: too close to call
A judge leans to a brand only when its confidence interval excludes 50%. Point estimates alone never count as a win.
The consensus gives equal weight to each model family. Per-judge results below are the primary measure.
Fast panel: the fast or small current models of each vendor, not the full flagship panel.
About this battle
- Category
- Midsize sedans
- As of
- Edition
2026-10- Methodology
ACM-1.0- Judges
- 4
- Scenarios
- 59
- Precision
- Measured
Results computed but not published
Only the overall consensus and per-judge results are published for this battle: each buyer need has fewer than 30 scenarios, the pre-registered minimum for a need row.
Per judge
95% confidence intervals (hull of the scenario-clustered bootstrap and Wilson intervals, widened for small samples in battles and newer boards)
| Judge (API model) | Share for Cadillac | Confidence interval | Leaning | Interval chart |
|---|---|---|---|---|
anthropic/claude-haiku-5.5-20261007 @ claude-haiku-5.5-20261007 via OpenRouter | 45% | 33%–58% | Too close to call | |
deepseek/deepseek-v4.1-flash-20260910 @ deepseek-v4.1-flash-20260910 via OpenRouter | 49% | 37%–62% | Too close to call | |
google/gemini-4-argon-20261013 @ gemini-4-argon-20261013 via OpenRouter | 64% | 52%–76% | Cadillac | |
openai/gpt-6-luna-20260922 @ gpt-6-luna-20260922 via OpenRouter | 50% | 37%–64% | Too close to call |
Who was counted
The rule that decides which judges count was fixed before the run. Judges that met it are counted with the weights below; the others are listed with their reasons.
- 59 buyer scenarios were planned.
- None of them was left out.
Consensus weights of 4 counted judges
| Judge (API model) | Model family | Weight in the consensus |
|---|---|---|
anthropic/claude-haiku-5.5-20261007 @ claude-haiku-5.5-20261007 via OpenRouter | anthropic | 25.0% |
deepseek/deepseek-v4.1-flash-20260910 @ deepseek-v4.1-flash-20260910 via OpenRouter | deepseek | 25.0% |
google/gemini-4-argon-20261013 @ gemini-4-argon-20261013 via OpenRouter | google | 25.0% |
openai/gpt-6-luna-20260922 @ gpt-6-luna-20260922 via OpenRouter | openai | 25.0% |
Judges not counted
jdg-live-mistral-large-4-0-20261006-or-v1- Too few valid answers (below the pre-registered minimum)
- Answered too few of the planned scenarios
Results computed but not published
Per-need scenario counts (n) are computed deterministically from the pre-registered plan.
- Buyer need: Daily commute
- n = 11 scenarios (fewer than 30) — Fewer scenarios than the pre-registered minimum
- Buyer need: Driving fun
- n = 13 scenarios (fewer than 30) — Fewer scenarios than the pre-registered minimum
- Buyer need: Family
- n = 14 scenarios (fewer than 30) — Fewer scenarios than the pre-registered minimum
- Buyer need: Reliability
- n = 12 scenarios (fewer than 30) — Fewer scenarios than the pre-registered minimum
- Buyer need: Low running costs
- n = 9 scenarios (fewer than 30) — Fewer scenarios than the pre-registered minimum
Methodology
Each judge is an AI model called through an API; “About this result” names each model and how it was called. In each buyer scenario, the judge chooses between the two brands. The scenario is the unit of analysis.
The results are based on 59 buyer scenarios.
A judge leans to a brand only when its confidence interval excludes 50%. Point estimates alone never count as a win.
Precision figures on this page are measured.
- Consensus, needs and languages
- 95% confidence intervals (MOVER combination of the per-judge intervals, widened for small samples in battles and newer boards)
- Per judge
- 95% confidence intervals (hull of the scenario-clustered bootstrap and Wilson intervals, widened for small samples in battles and newer boards)
- Scenario languages
- English
Data warnings
- Models chose on EPA specification facts only; purchase price, reliability and availability were not assessed.
About this result
What was measured, how, when, and how to check it.
- Methodology
ACM-1.0- Edition
2026-10- Measured on
- AI models (API)
anthropic/claude-haiku-5.5-20261007 @ claude-haiku-5.5-20261007 via OpenRouter, via OpenRouter (pinned providers)deepseek/deepseek-v4.1-flash-20260910 @ deepseek-v4.1-flash-20260910 via OpenRouter, via OpenRouter (pinned providers)google/gemini-4-argon-20261013 @ gemini-4-argon-20261013 via OpenRouter, via OpenRouter (pinned providers)openai/gpt-6-luna-20260922 @ gpt-6-luna-20260922 via OpenRouter, via OpenRouter (pinned providers)
- Conditions
- Each model was called through the route listed next to it and chose between the two brands in 59 buyer scenarios. Scenario languages: English.
- Judges counted
The pre-registered inclusion rule counted 4 of the 5 judges on the panel. The results use 59 of the 59 planned scenarios.
- Pre-registration (SHA-256)
40514d9cdea6c4c742e4ee6e79217c2ae693d8150ab9991e5ea5760b59ea7b74- What the pre-registration fixes
- Signed snapshot
Verify this result: download the signed snapshot (JSON)
- Signing key (kid)
FnG9tqyrr6Y0YUh6teXhpERjAKUSfzL78m4WdnEZ9Jg- Production key
- SHA-256 of the payload
54c89f5a32ad2b29571f4537466fb9560ad639eb978dae47fb8dcfe6c655e5fa
Rank, inclusion and influence on results are never sold.