On the field
Where each model's cloud of answers settles, and the gap between them.
Llama
ClaudeCharacter
How far each leans, how steadily it holds, how far it bends under pressure, and how often it answers.
LlamaClaude The takeaway
Claude and Llama both score -0.06 on the economic left-right axis, placing them at Center. Their overall political leans are indistinguishable at this sample size, as the 95% confidence intervals overlap.
The largest disagreement is on same-sex marriage: Claude strongly supports (0.69) while Llama is balanced (0.05). On encryption backdoors, Claude clearly opposes (-0.42) while Llama is balanced (-0.03). They fully agree on four economic policies, all scoring 0.0 (balanced).
Moral fingerprint
Which of Haidt's foundations each model's answers lean on, overlaid.
LlamaClaudeWhere they most disagree
The questions with the widest gap between the two stances. Open a row to read both answers.
LlamaClaudeCommon ground
Where the two land in close agreement.
LlamaClaudeThis diffs both models on their raw weights (Condition A). Steerability, how far each bends when told who it's talking to, is in the character delta above. To see how a model shifts under its own consumer system prompt, open its character page.
Common questions
Is Claude more left-wing than Llama?
No. Both have identical economic lean values of -0.06 (Center), so they are not distinguishable on the left-right axis.
Where do Claude and Llama agree?
They agree on wealth tax over $50M, cutting corporate taxes, less business regulation, privatizing public services, and cut spending over stimulus, all scoring 0.0 (balanced).
Which of Claude and Llama is more consistent?
Llama has a higher stability percentage (88% vs Claude's 82%), indicating more consistent responses across repeated prompts.
Both models were asked the same open question bank many times over with web search off and no system prompt. Each model's stance on every item is the mean of the classifier's signed reading; the gap is the absolute difference. "Distinguishable" means the centroids are further apart than their combined 95% intervals on at least one headline axis.
