ModelsThe Decoder· Jun 30, 2026
Anthropic Releases Claude Sonnet 5, Narrowing Gap with Opus Models
Anthropic has introduced Claude Sonnet 5, an updated AI model that outperforms its predecessor across benchmarks. The new model also exceeds the larger Opus 4.8 on a knowledge work evaluation. Anthropic notes its lower scores on cybersecurity tasks relative to certain restricted models.
Key points
- →Claude Sonnet 5 beats Sonnet 4.6 on all benchmarks
- →Edges past Opus 4.8 on GDPval-AA v2 test
- →Scores below US-blocked models on cybersecurity tasks
Mentioned
AnthropicClaude Sonnet 5Claude Sonnet 4.6Claude Opus 4.8
Related stories
Google Launches Nano Banana 2 Lite as Fastest Cheapest Image ModelArs Technica · Models→Anthropic Releases Claude Science AI Workbench for ResearchersThe Decoder · Product→Anthropic Releases Claude Sonnet 5 for Cost-Effective Agent UseTechCrunch · Models→Google Launches Nano Banana 2 Lite and Gemini Omni Flash AI ModelsThe Decoder · Models→Anthropic Introduces Claude Science Workbench for ResearchersTechCrunch · Tools→Meituan Trains 1.6 Trillion Parameter Model on Chinese ChipsThe Decoder · Models→
This is an original summary by Dhanasvi's agents based on The Decoder's public feed. For the complete article, visit the original source. Trademarks and article copyright belong to their owners.