Anthropic ships Claude 4 with extended thinking and 1M token context
Anthropic released Claude 4 Opus, featuring a new "extended thinking" mode that lets the model reason through complex problems before answering. The 1M token context window makes it the first model capable of ingesting an entire codebase in a single prompt.
Anthropic released Claude 4 Opus today, marking a significant leap in large language model capabilities. The flagship release introduces two headline features: an "extended thinking" mode and a 1-million-token context window.
Extended Thinking Mode
Extended thinking allows Claude 4 to break complex problems into a visible chain of reasoning before producing a final answer. Internal benchmarks show double-digit improvements on math, code, and multi-step planning tasks compared with Claude 3.7.
1M Token Context
The expanded context window puts Claude 4 ahead of GPT-5 and Gemini Ultra 2 on long-document workloads. Early enterprise testers report ingesting full monorepos, multi-thousand-page contracts, and complete medical histories in a single prompt without retrieval.
Pricing & Availability
Claude 4 Opus is available today in the API at $12 / 1M input tokens and $60 / 1M output tokens, with extended thinking billed separately. A consumer Pro tier launches next week at $25/month.