新用戶 免費註冊,送 10 次呼叫,最高 $1,免綁卡。

Claude Opus 5.5

發布於 2026-09-22

chat程式碼推理工具呼叫影像理解提示詞快取

Claude Opus 5.5 是 Anthropic 針對長時間執行的智慧體編碼與知識工作推出的模型,於 2026 年 9 月 22 日發布;Anthropic 的模型總覽建議多數工作負載從這裡起步,把 Claude Fable 5.1 留給最嚴苛的推理任務。

輸入
文字 影像 $4/M
輸出
文字 $20/M
快取讀取
$0.2/M
上下文
1M
相較 GPT-4o
便宜約 20%
知識截止
2026-06

Benchmark 成績

高於同儕均值無人分數更高9 / 97 / 9
Claude Opus 5.5 其他被測模型 同儕均值 無人分數更高
Terminal-bench 4.0
無人分數更高 66.4%
OSWorld 2.0 partial
無人分數更高 81.8%
Terminal-Bench-Science 0.1
58.7%
Humanity's Last Exam with tools
無人分數更高 67.7%
AutomationBench
40%
Chartography with tools
無人分數更高 89%

廠商公布: Alibaba (Qwen) Anthropic DeepSeek Google Moonshot OpenAI Z.ai

價格在同類中的位置

價格在 67 個同類模型中的位置

輸入$4/M
$0.05 · Qwen3 VL Flash GPT-5.4 Pro · $30
輸出$20/M
$0.275 · DeepSeek V4 Flash GPT-5.4 Pro · $180
快取讀取$0.2/M
$0.0028 · DeepSeek V4 Flash GPT-5.4 Pro · $15

這條線顯示該模型的價格,在 Synthorai 上同類模型裡處於什麼位置。兩端標出了最便宜和最貴的那個。這裡是基礎價,批次、區域與快取寫入的折扣見價格頁。

規格與限制

Token

上下文視窗(廠商規格) 1,000,000
最大輸出(廠商規格) 128,000
知識截止 2026-06

提示詞快取

快取方式 顯式(需開啟)
最低前綴 512 廠商預設 1,024
存活時間 預設 5 分鐘,可選 1 小時
寫入成本 1.25x (5m) / 2x (1h)

思考

廠商參數 output_config.effort
可選值 low · medium · high · xhigh · max
預設值 medium 請求未指定時生效
可關閉 不支援
思考行為 Adaptive thinking is always on: thinking {"type": "disabled"} and a manual {"type": "enabled", "budget_tokens": N} both return a 400 error, so effort is the only depth control. The default dropped from high on Claude Opus 5 to medium.
參數 reasoning_effort
取值 minimal · low · medium · high 閘道側參數面——以上方廠商映射為準

模型

模態 文字 + 影像 → 文字
  • 1M context at standard pricing with no long-context tier
  • prompt-cache reads cost 0.05x input ($0.20/M) rather than the usual 0.1x
  • minimum cacheable prompt 512 tokens
  • forced tool use (tool_choice any or tool) returns a 400
  • thinking blocks are tied to the model and conversation that produced them
  • up to 300k output tokens on the Message Batches API with the output-300k-2026-03-24 beta header
  • fast mode is a research preview on the Claude API only

依據 Anthropic 官方文件 ↗

單一提示詞 — 經由閘道測量

提示詞 Explain to a CFO, in exactly three sentences, why our LLM bill doubled after we enabled reasoning. 檢查 恰好 3 句

Claude Opus 5.5 通過 · 3 sentences

When we turned on reasoning, the model began "thinking out loud" before each answer, generating thousands of internal working tokens that users never see but that we are billed for just like the final response. Those hidden tokens are charged at the output rate, typically the most expensive rate, so the same question with the same visible answer now costs roughly twice as much or more. We can bring the bill back down without losing the quality gains by capping the reasoning budget and reserving reasoning mode for the complex tasks that actually benefit from it.

輸出 263 tok (+102 思考) 延遲 5.7 s

指令遵循(恰好三句,可數)、受眾適配(面向 CFO 的語氣),以及下方 token 計量所暴露的隱藏思考計費缺口。

提示詞 This function is supposed to return the longest run of consecutive equal items, but callers report it is off by one on single-element inputs and misses runs that end at the last index. Fix it and explain the bug in one paragraph. 檢查 修復通過測試

Claude Opus 5.5 通過 · 8/8 cases

```python def longest_run(xs): if not xs: return 0 best = 1 cur = 1 for i in range(1, len(xs)): if xs[i] == xs[i - 1]: cur += 1 best = max(best, cur) else: cur = 1 return best ``` **Explanation:** The original code counted *matching adjacent pairs* rather than *items in a run*. Both `cur` and `best` started at 0, and `cur` reset to 0 on a mismatch, so a run of length *n* was reported as *n − 1*. For a single-element list the loop never executes, so it returned 0 instead of 1. For `[5, 5]` it returned 1 instead of 2. The fix treats every element as the start of a run of length 1: `best` and `cur` start at 1, and `cur` resets to 1 (not 0) when the value changes. An explicit check returns 0 for an empty list. The "misses runs at the last index" reports are the same off-by-one bug rather than a separate one. […]

輸出 723 tok (+303 思考) 延遲 10.4 s

修復是否真的正確(可執行)、解釋的資訊密度,以及在一個邊界明確的任務上的 token 效率。

提示詞 Extract a JSON object with fields {company, quarter, revenue_usd_m, growth_pct, guidance} from this text. Use null for anything not stated; add no other fields. 檢查 合法 JSON,schema 精確

Claude Opus 5.5 通過 · 5/5 fields, guidance "Withheld pending Q4 close"

```json { "company": "Meridian Robotics", "quarter": "September quarter", "revenue_usd_m": 412, "growth_pct": 11, "guidance": "Withheld pending Q4 close" } ``` **Judgment calls to check:** - **quarter**: Kept as "September quarter" rather than "Q3", since the fiscal calendar isn't stated. The text mentions an upcoming "Q4 close," so it may be fiscal Q3, but that's not certain. - **revenue_usd_m**: The text says only "412 for the topline." It doesn't state currency or units, and the figure is on a constant-currency basis, not reported USD. USD millions is an assumption. Use `null` if you need strictly stated values. - **growth_pct**: The source says "eleven and change," so the true figure is slightly above 11. 11 is an approximation. - **guidance**: Guidance was explicitly withheld. I recorded that rather than `null`, since the text addresses it directly.

輸出 829 tok (+488 思考) 延遲 10.9 s

schema 服從度(不臆造欄位)、幻覺壓力(guidance 明確被暫緩給出),以及結構化輸出路徑的差異。

提示詞 Write a 120-word product announcement for an API gateway feature that caches prompts across providers. Forbidden words: "seamless", "unlock", "game-changer", "revolutionize", "empower". Exactly one sentence must be a question. 檢查 120 詞,0 個禁用詞

Claude Opus 5.5 通過 · 120 words, 0 banned, 1 question

Today we're launching Cross-Provider Prompt Caching for our API gateway. Why pay twice for the same context? Our gateway now stores repeated prompt prefixes, system instructions, and document chunks, then reuses them whether your request goes to OpenAI, Anthropic, Google, or an open-source model. Teams running retrieval pipelines or long agent loops can cut token costs by up to 60 percent and trim latency on every cached call. Setup takes one configuration flag; no code changes are required. Cache rules respect tenant boundaries, expire on your schedule, and log every hit for audit review. Dashboards show savings by provider, route, and application in real time. Cross-Provider Prompt Caching is available on all Pro and Enterprise plans. Turn it on today.

輸出 1263 tok (+1007 思考) 延遲 13.5 s

約束服從度(字數預算、禁用詞表、唯一的那句問句)、文風指紋,以及長度控制。

30 秒用上 Claude Opus 5.5

OpenAI 相容:換掉 base_url,SDK 不用改。POST /v1/chat/completions

from openai import OpenAI

client = OpenAI(
    base_url="https://synthorai.io/v1",
    api_key="sk-syn-...",
)

resp = client.chat.completions.create(
    model="claude-opus-5-5",
    messages=[{"role": "user", "content": "Summarize this diff"}],
    reasoning_effort="medium",
)
print(resp.choices[0].message.content)

關於 Claude Opus 5.5

  • 它的價格為每百萬輸入 token $4、每百萬輸出 token $20,提示詞快取讀取按輸入價的 5%(每百萬 $0.20)計費,而非一般的 10%,因此反覆重放穩定前綴的智慧體迴圈會明顯更便宜。
  • 它保留 1M token 上下文視窗且沒有長上下文加價,最大輸出 128K,支援文字與影像輸入,知識截止為 2026 年 6 月。
  • Anthropic 文件列出從 Claude Opus 5 遷移時的四項破壞性變更:自適應思考始終開啟,關閉思考或設定手動預算會回傳錯誤,預設改為 medium 的 effort 參數成為唯一的深度控制;透過 tool_choice any 或 tool 強制呼叫工具會被拒絕,需保持 tool_choice 為 auto,並以嚴格工具呼叫或結構化輸出取得符合 schema 的 JSON;思考區塊與產生它的模型和對話綁定;Claude API 不再接受較早的 computer_20251124 工具。
  • 模型在工具呼叫之間寫下的簡短說明現在以思考區塊回傳,需要串流顯示這些內容的介面應設定 thinking 的 display 值。
  • Synthorai 透過與其餘 Claude 模型相同的 API 提供 Claude Opus 5.5。

常見問題

Claude Opus 5.5 API 可以免費試用嗎?

可以,新帳號可獲得 10 次試用呼叫和最高 $1 的免費額度,無需信用卡。以輸入 $4/M 計算,光是這筆額度就足以對 Claude Opus 5.5 發出約 31 次 ~8K token 的請求。

Claude Opus 5.5 最擅長什麼?

Anthropic 建議多數工作負載的起步模型、快取讀取為輸入價的 5%,每百萬 token $0.20、思考始終開啟,力度預設 medium。完整能力請見「關於」一節,內容取自廠商官方發布說明。

Claude Opus 5.5 的價格是多少?

在 Synthorai 上,Claude Opus 5.5 輸入 $4/百萬 token、輸出 $20/百萬 token,即廠商牌價,無平台加價。快取命中的輸入 token 以 $0.2/M 計費。

Claude Opus 5.5 支援提示詞快取(prompt caching)嗎?

支援,需主動開啟:以 cache_control 中斷點標記穩定前綴。快取命中的輸入 token 以 $0.2/M 計費(未命中 $4/M);提示詞需有 512 個 token 以上的穩定前綴才能命中快取(TTL 預設 5 分鐘,可選 1 小時)。 提示詞快取指南 →

如何開通 Claude Opus 5.5?

把現有 OpenAI SDK 的 base_url 指向 "https://synthorai.io/v1",model 設為 "claude-opus-5-5" 即可。一組 API key 通用閘道上的所有模型。

Claude Opus 5.5 的知識截止日期是什麼時候?

Claude Opus 5.5 的知識截止日期為 2026-06,依據廠商官方文件(資料核驗於 2026-09-23)。

相關模型

對比

本頁每個值都轉錄自廠商自己的文件(連結見上),並帶有核對日期。價格在全目錄範圍內比較;各廠商定義不同的規格值,只說明差異而不作圖表對比。此處沒有任何由我們測量的資料,也不做評分。

取得 API 金鑰 算算你的成本 →