Gate News, on March 14, Anthropic announced that Claude Opus 4.6 and Claude Sonnet 4.6 are now officially fully available, with both models supporting a 1 million Token context window. This capability was previously in Beta phase and is now available to users with pricing remaining unchanged. Regarding pricing, Opus 4.6 maintains 5 dollars per million input Tokens and 25 dollars for output, while Sonnet 4.6 is 3 dollars and 15 dollars respectively, with consistent unit pricing from the first to the millionth Token. In the long context capability test MRCR v2, Opus 4.6 achieved 78.3% accuracy, surpassing GPT-5.4's 36.6% and Gemini 3.1 Pro's 25.9%. Additionally, the new models have enabled million-token context in Claude Code, which can significantly reduce context compression issues during large codebase analysis and log retrieval, and support up to 600 images or 600 pages of PDF in a single request.