0.81
- 開始
- 2025年11月24日確度 95%出典：anthropic.com
Claude Opus 4.5 Released
Anthropic releases Claude Opus 4.5, its new flagship model with state-of-the-art coding and agentic performance at lower pricing.
- Anthropic announced Claude Opus 4.5 on November 24, 2025, available on its apps, the API (claude-opus-4-5-20251101) and all three major cloud platforms[1]
- Priced at $5 per million input and $25 per million output tokens[1][3]
- Product updates: Plan Mode in Claude Code, Claude Code in the desktop app, automatic summarizing of long conversations, Claude for Chrome for all Max users, and higher usage limits for Max and Team Premium[1]
- The Opus product page lists Opus 4.5 (Nov 24, 2025) as the start of the Opus 4.5 to 4.8 line, later followed by Opus 5[3]
Notable features
- Described as the best model in the world for coding, agents and computer use, and it scored higher than any human candidate on Anthropic's 2-hour performance engineering take-home[1]
- It has a 200K-token context window and up to 64K output tokens, takes text and image input, supports extended thinking with a default effort of high, and has a reliable knowledge cutoff of May 2025[2]
- Benchmarks: leads 7 of 8 languages on SWE-bench Multilingual and gains 10.6% over Sonnet 4.5 on Aider Polyglot[1]
- New effort parameter: at medium effort it matches Sonnet 4.5's best SWE-bench Verified score with 76% fewer output tokens[1]
- Safety: called the most robustly aligned model Anthropic had released, and harder to trick with prompt injection than other frontier models[1]
参考資料 3確度 90%
最初の項目は常にピンの出典です。全体の確度は、各資料が上記の開始・終了時刻をどれだけ強く裏付けているかの加重平均です。最新の資料より180日古くなるごとに、重みは半分になります。
- [1]95%Claude Opus 4.5anthropic.com· 投稿 2026年9月19日· 開始 2025年11月24日· スコアの 33%
Anthropic's[3] own announcement, dated the day the model became available.
- [2]90%Claude Opus 4.5 - Claude Platform Docsplatform.claude.com· 追加 2026年9月22日· スコアの 34%
Anthropic's[1][3] model reference page for Claude Opus 4.5 gives "Context window: 200K tokens · Max output: 64K tokens", text and images to text, extended thinking with default effort high, and a May 2025 reliable knowledge cutoff.
- [3]85%Claude Opus \ Anthropicanthropic.com· 追加 2026年9月19日· 開始 2025年11月24日· スコアの 33%
Anthropic's[1] Claude[2] Opus page lists this announcement and its date.
修正を提案
足りない点や誤りがありますか？ このピンを裏付けるリンク、別の開始日・終了日とその理由、欠けている情報や誤っている情報を、自由に書いてください。AIがこのピンの出典と照らし合わせ、より良い情報源を探し、裏付けとなるページがあれば参考資料として追加します。ピン自身の出典が引き続き最も重視されます。
