Top News

In the world of artificial intelligence, Anthropic has introduced a major update. The company has released the Claude Opus 4.6 AI model, which is said to be more intelligent, more stable, and capable of handling longer tasks compared to previous versions. Importantly, this model can now process very large amounts of data and long conversations without a drop in performance. It has been specifically strengthened for software engineering, research, and knowledge-based tasks. The company claims that its capabilities in agentic workflows and coding tasks have significantly improved.

The Power of Long-Context Understanding with 1 Million Tokens
The biggest upgrade in Claude Opus 4.6 is its one-million-token context window. Previous Opus models were limited to approximately 200,000 tokens, which led to performance degradation with long files, large databases, or extensive conversations. In the new beta version, this limit has been increased fivefold. This means the model can understand a much larger amount of text and data simultaneously. It will provide more stable responses in long research projects, legal documents, technical repositories, and multi-step analyses. The company has specifically optimized it for long-horizon tasks.

A New Level of Agentic Workflows and Adaptive Thinking
This model incorporates new techniques such as context compaction, which summarizes older data during long tasks to keep the system updated. It also features adaptive thinking, where the model determines the depth of its reasoning based on the complexity of the question. Users can set different effort controls to balance speed, intelligence, and cost. In agent-based tasks, this model can plan multiple steps independently to advance the task. This makes automation and research workflows more efficient.

Strong Performance in Coding, Research, and Business Tasks
According to Anthropic, Claude Opus 4.6 can handle large code repositories independently, perform code reviews, and demonstrate improved accuracy in debugging. Through the Claude Code Research Preview, it can also create agent teams for collaborative development. In internal benchmark tests, it was found to be on par with or superior to top models in command-line and multi-disciplinary reasoning. In agentic evaluations across domains such as finance and legal, it scored better than previous versions and several competing models. Its performance in areas like computational biology is also reported to be nearly double.

Comprehensive Security, Safety, and Availability Information
The company states that safety has been prioritized in the new model. Additional testing and audits have been conducted to mitigate risks such as deception and over-reliance. Six new cybersecurity probes have been added to help identify potential misuse. Claude Opus 4.6 is available on the website, mobile and desktop apps, API, and major cloud platforms. API pricing starts at $5 per million input tokens and $25 per million output tokens, with a separate premium applicable for larger contexts. This provides both developers and enterprise users with direct access to advanced AI tools.

Disclaimer: This content has been sourced and edited from TV9. While we have made modifications for clarity and presentation, the original content belongs to its respective authors and website. We do not claim ownership of the content.