Claude Opus 4.6

Claude Opus 4.6, released on February 5, 2026, is one of Anthropic's most capable production models. It is the first Opus to bring a million-token context window and native multi-agent collaboration (Agent Teams) to the lineup, and it doubles max output tokens to 128K, so a single response can deliver a complete code module, a multi-file refactor, or a long-form research report without truncation.

Core Capabilities 1M context window: Around 2,500 pages of text — analyze entire codebases or long documents end-to-end (GA with unified pricing as of March 2026). 128K max output: Doubled from the previous generation, ideal for very long single-shot generations. Adaptive Thinking: The model evaluates each request and engages deep reasoning on hard problems while skipping it on simple prompts to save latency and cost. Native multi-agent collaboration: Multiple Opus 4.6 instances can form Agent Teams to tackle long-horizon tasks that weren't previously feasible. Top-tier benchmarks: 80.9% on GPQA Diamond (graduate-level science reasoning), 80.8% on SWE-bench Verified. Technical Specifications Model ID: claude-opus-4-6 Context window: 1,000,000 tokens Max output: 128,000 tokens (extendable to 300K on the Batch API via a beta header) Multimodal: Text + image input, text output Pricing: $15/M input tokens, $75/M output tokens (see the official Pricing page for current rates) Typical Use Cases Code tasks that need to emit an entire module or multi-file refactor in one shot Research reports, white papers, and technical docs longer than 50,000 words Cross-document synthesis and comparison across multiple long sources Complex business process automation with multi-agent coordination Deep reasoning in scientific, legal, and financial domains Recommendations If you want Opus-level intelligence with more predictable cost and behavior than 4.7, Opus 4.6 remains an excellent choice — especially for workflows already running stably in production. For new projects, evaluate whether Opus 4.7's agentic gains justify the migration effort.

Available On Claude API, Claude Platform on AWS, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.