BREAKING
Claude Top Model Costs Up to 5x More
Claude Output Price per 1M Tokens
Haiku 4.5
5
Sonnet 4.6
15
Opus 4.8
25
SWE-bench Verified Gap Is Slim
Haiku 4.5
73.3
Sonnet 4.6
79.6
Opus 4.8
82
Cheaper Default vs Premium Depth
Cheaper models
Default
●
Enough for everyday tasks
●
Faster and lower cost
●
Best for high-volume queries
Top-tier Opus
Depth
●
Complex reasoning
●
Long agent runs
●
Fewer reworks
Model Routing by Task
1
Assess task complexity
↓
2
Simple to cheaper model
↓
3
High-stakes to premium
Price Gap Outweighs Skill Gap
AI NEWS BLITZ
Anthropic's flagship Claude can cost up to five times more, sparking a model-choice debate.