Claude Opus 4
Claude Opus 4 is a high-performance large language model in Anthropic's Claude 4 series, representing the flagship tier of that lineup. It possesses strong natural language understanding and generation capabilities, excels at complex reasoning, long-form text processing, and multi-step task execution, and demonstrates notable performance in technical tasks such as code writing and front-end interface generation. It is designed for professional use cases requiring high levels of reasoning depth and task complexity.
Core Facts
Timeline (last 90 days)
Bug two was that a configuration error in Claude API's TPU servers caused incorrect token generation, occasionally assigning high probability to tokens that should rarely appear, resulting in English prompts getting Thai or Chinese replies and code syntax errors; this problem affected Opus 4.1 and Opus 4 from August 25 to 28 and Sonnet 4 until a rollback on September 2, and third-party platforms were not affected.
During the tutorial's verification step, the model name had expired on the first run, requiring a switch from Claude Opus 4 to Opus 3.5 for the call to work properly.
The 11 most well-known writing tics lost 53% of their strength from Opus 4 to Opus 5.5, while over the same period the total number of tics dropped from 3,746 to 2,548, a decrease of only 32%.
Chroma's 2025 research tested 18 frontier models (including GPT-4.1, Claude Opus 4 and Gemini 2.5) and found that accuracy did not decline uniformly as input length grew, and that even a single distractor paragraph noticeably harmed performance relative to a cleaner baseline.
Anthropic released Claude Opus 5, with significant improvements over Opus 4 in coding, visual understanding, and agent coordination.
All Facts (10)
The Claude Opus 4 series is Anthropic's flagship large language model series launched in 2025
90%UnverifiedClaude Opus 4和Sonnet 4都被定义为混合模型（Hybrid Models），将快速响应模式与深度推理模式整合在同一模型架构中
90%UnverifiedOpus 4在编程实测中能精准地删除和修改需要变更的代码部分，而不是从头重写
85%UnverifiedClaude Opus 4属于推理增强型模型，处理复杂任务时会启动扩展思维链（Extended Thinking），单次复杂推理的算力成本可达普通模型的20-50倍
70%UnverifiedThe 11 most well-known writing tics lost 53% of their strength from Opus 4 to Opus 5.5, while over the same period the total number of tics dropped from 3,746 to 2,548, a decrease of only 32%.
50%UnverifiedChroma's 2025 research tested 18 frontier models (including GPT-4.1, Claude Opus 4 and Gemini 2.5) and found that accuracy did not decline uniformly as input length grew, and that even a single distractor paragraph noticeably harmed performance relative to a cleaner baseline.
50%UnverifiedClaude Opus 4这类顶级模型参数量可能超过数千亿
50%UnverifiedBug two was that a configuration error in Claude API's TPU servers caused incorrect token generation, occasionally assigning high probability to tokens that should rarely appear, resulting in English prompts getting Thai or Chinese replies and code syntax errors; this problem affected Opus 4.1 and Opus 4 from August 25 to 28 and Sonnet 4 until a rollback on September 2, and third-party platforms were not affected.
50%UnverifiedDuring the tutorial's verification step, the model name had expired on the first run, requiring a switch from Claude Opus 4 to Opus 3.5 for the call to work properly.
50%UnverifiedAnthropic released Claude Opus 5, with significant improvements over Opus 4 in coding, visual understanding, and agent coordination.
50%