BREAKING
Anthropic Ships Claude Sonnet 5
92.4% on SWE-bench Verified
Sonnet 592.4
Opus 4.680.8
Gemini 3.1 Pro80.6
GPT-5.457.7
0%
OSWorld
0%
GPQA Diamond
0%
ARC-AGI-2
0$
input / Mtok
0$
output / Mtok
0M
context window
Praise and Pushback
Praise
●Frontier perf at Sonnet price
●Stable agentic work
●Claude Code and Devin
Pushback
●Some scores below Sonnet 4.6
●Trails Opus 4.8
●Talk of benchnerfing
Mid-Tier Goes Frontier
AI NEWS BLITZ
Anthropic has officially launched Claude Sonnet 5, its new mid-tier model.