Claude Sonnet 4.5
baselineBaseline/reference model used for comparison. Strong natural formatting abilities due to fine-tuning and training data.
Pricing
Input Cost
$3/Million Tokens
Output Cost
$15/Million Tokens
Strengths
- Natural markdown formatting
- Strong instruction following
- Excellent prose generation
- Consistent formatting without explicit instructions
Weaknesses
- High cost ($3/M input, $15/M output)
- Rate limits at scale
- Occasional issues with answer keys in assessments
Analysis
Claude Sonnet 4.5 serves as the gold standard reference model. Its system prompt has strong influence on formatting, and it naturally provides bullet points, prose, and structured content without explicit instructions. However, cost and rate limits make it unsuitable for production at scale.