BREAKING
Anthropic Launches Claude Sonnet 5
Opus-Level Skills at Sonnet Prices
Agentic Coding Benchmark
Sonnet 4.6
58.1
Sonnet 5
63.2
Opus 4.8
69.2
0
$
Input per M
0
$
Output per M
0
$
Standard output
Strengths vs Limitations
Strengths
●
Completes complex multi-step tasks
●
Self-verifies its own outputs
●
End-to-end agent automation
Limitations
●
Mixed views on naturalness
●
0% on Firefox 147 exploit eval
●
Opus 4.8 still leads hardest tasks
A New Default for Coding Work
AI NEWS BLITZ
Anthropic has released Claude Sonnet 5, its most agentic mid-tier model yet.