When you purchase through links on our site, we may earn an affiliate commission. This doesn’t affect our editorial independence.
Anthropic has officially launched Claude Opus 4.6, the latest of its most advanced AI models. This update is specifically designed to enhance performance in coding workflows and to handle exceptionally long documents, marked notably by the introduction of a beta 1-million-token context window, a first for the Opus class of models.
The model is now accessible through Anthropic’s main chat interface, its API, and major cloud platforms. Developers can access it via the API identifier claude-opus-4-6. Pricing for standard use remains at $5 per million tokens for input and $25 per million for output, though a premium rate applies to prompts exceeding 200,000 tokens.
Core Features of Claude Opus 4.6
In terms of core capabilities, Claude Opus 4.6 builds upon its predecessor with significant improvements in coding tasks. Anthropic states it demonstrates more careful planning, better execution of long-running, multi-step “agentic” tasks, and greater reliability when working within large codebases. It also features stronger code review and self-debugging features.
These technical enhancements are designed to benefit broader knowledge work beyond coding. Anthropic highlights applications in financial analysis, research, and the creation or manipulation of documents, spreadsheets, and presentations. These complex workflows can be executed within Claude’s Cowork environment, where the model can operate autonomously.
Anthropic supports its performance claims with several benchmark results. The company reports Opus 4.6 achieves top scores on Terminal-Bench 2.0 for agentic coding and leads on Humanity’s Last Exam for multidisciplinary reasoning. It also claims a significant Elo point advantage over competitors on GDPval-AA, a test of economically valuable knowledge work.
Anthropic Rolls Out Platform Updates
Accompanying the model release are key platform updates to support complex, long-horizon tasks. In Claude Code, users can now assemble teams of AI agents to collaborate. The Developer Platform introduces “adaptive thinking,” allowing the model to autonomously decide when to use extended reasoning.
Further platform controls include adjustable “effort” settings—low, medium, high, and max—letting developers balance intelligence, speed, and cost. To manage the new long-context capabilities, a beta “context compaction” feature on the API summarises older parts of a conversation to stay within token limits, enabling longer-running interactions.
The release also brings major upgrades to Anthropic’s office software integrations. Substantial improvements have been made to Claude in Excel, and the company is launching a research preview of Claude in PowerPoint. This integration allows the AI to edit slides while respecting templates and can generate entire presentations from a description.
Anthropic emphasises the safety profile of Claude Opus 4.6, stating that its system card shows it has an overall safety standard as good as, or better than, that of other leading models, with low rates of misaligned behaviour in evaluations. GitHub’s chief product officer offered positive early feedback, noting its promise for complex, multi-step developer work.
This launch occurs amidst significant growth for Anthropic, following reports of plans to raise substantial new funding of up to $10B. Also, in December 2025, it entered a $200M deal with Snowflake. The company aims to sharply increase revenue this year, driven by the adoption of enterprise products such as the newly enhanced Claude Opus 4.6, which consolidates its focus on complex, professional-grade AI applications.
