Anthropic officially launched its Claude Sonnet 5 model on February 3, 2026, marking a significant advancement in AI capabilities, particularly for coding tasks. The model, internally codenamed "Fennec," achieved an 82.1% score on the demanding SWE-Bench Verified benchmark, surpassing previous models including Anthropic's own Opus 4.5 and OpenAI's estimated GPT-5 performance. This release positions Sonnet 5 as a formidable competitor in the rapidly evolving AI landscape.
A key feature of Claude Sonnet 5 is its expansive 1-million-token context window, a fivefold increase over Opus 4.5. This allows the model to process entire codebases and maintain coherent understanding across numerous files, facilitating complex refactoring and analysis. The model is also optimized for agentic workflows, capable of proactively managing multi-step tasks, executing code in a built-in terminal, and self-correcting errors.
Perhaps the most disruptive aspect of Sonnet 5 is its pricing structure, offering an approximate 80% cost reduction compared to Opus 4.5. With input priced at $3.00 and output at $15.00 per 1 million tokens, it delivers superior performance at a significantly lower cost. This cost-efficiency makes it particularly appealing for enterprise and high-volume coding workloads, challenging the market positions of other leading AI developers.
The model is available through the Anthropic API, Claude Pro, and Google Vertex AI, indicating broad accessibility for developers and businesses. Its optimization for Google’s Antigravity TPU infrastructure further enhances its inference speed and efficiency. While the tweet suggested a release "next week," Sonnet 5 has been actively deployed since early February, already impacting the competitive landscape.
