Benchmark performance

Mashable • February 18, 2026 • business

Key Points:

Anthropic has released Claude Sonnet 4.6, its most capable Sonnet model to date, featuring a 1 million token context window in beta and improved coding skills, available now for free and Pro users on claude.ai, Claude Cowork, API, and major cloud platforms.
Claude Sonnet 4.6 demonstrated strong performance on internal safety tests, showing low hallucination and sycophancy tendencies, making it a reliable option for developers and users.
Benchmark tests reveal that Claude Sonnet 4.6 outperforms competitors like Google's Gemini 3 Pro and OpenAI's GPT 5.2 in financial analysis and office tasks, even surpassing Anthropic's own Opus 4.6 model in