Anthropic's release of Claude Opus 5 with a 1-million-token context window pushes the boundaries of what AI models can process in a single interaction. The window is large enough to ingest an entire codebase, a full legal case file, or years of financial records. [1]

The context window race has accelerated throughout 2026. Google's Gemini offered 2 million tokens earlier this year. OpenAI's GPT-5.6 supports 500,000. Anthropic's entry at 1 million positions it between the two, with the company emphasizing quality of retrieval over raw capacity. [1]

The practical implications are significant. Legal firms can upload complete case histories. Software teams can feed entire repositories for code review. Researchers can process entire literature collections. The constraint shifts from "what fits in the context window" to "what questions to ask of the full corpus." [2]

Anthropic's claim is that Opus 5 maintains accuracy across the full million tokens — a technical challenge that has plagued competing models. Long-context performance degrades as the window expands, with models losing track of information buried in early positions. [2]

The release arrives amid OpenAI's aggressive pricing strategy. Anthropic's positioning emphasizes capability over cost — a bet that enterprises will pay more for reliable long-context performance. [1]

-- KENJI NAKAMURA, Tokyo