Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Claude Fable 5.1 和 Claude Mythos 5.1 的行为差异与提示模式，涵盖 effort、进度更新、工具调用批处理、对话历史、写作风格、格式、任务完成、压缩摘要、范围与测试覆盖、搜索触发、安全防护误报、文件编辑、长输出、子智能体以及视觉。
有关该模型的能力、API 变更、定价和可用性，请参阅 Claude Fable 5.1 的新功能。有关适用于所有 Claude 模型的技巧，请参阅提示最佳实践。
您现有的 Claude Fable 5 提示无需修改即可在 Claude Fable 5.1 上良好运行，但有少数行为差异值得了解。请从与您观察到的现象相符的章节开始：
bound to a different conversation 而失败，或者您的 harness 在请求之间编辑了较早的轮次：保持对话历史仅追加stop_reason: "refusal"：减少安全防护误报xhigh 或 max effort 下的长篇交付物耗时很长或触及 max_tokens：在 xhigh 和 max effort 下为长输出留出空间从默认的 effort（努力程度）级别 high 开始，然后针对您自己的评估测试其他级别（low、medium、xhigh 和 max）。在 Claude Fable 5.1 上，effort 是在智能、延迟和成本之间进行权衡的主要控制手段。即使您已经在 Claude Fable 5 上做过一次扫描测试，也请重新运行：effort 级别名称在不同模型之间并不对应相同的思考量。
Claude Fable 5.1 相对于 Claude Fable 5 的能力提升在各个 effort 级别上都有体现，并且在较高设置下最为显著。在 medium 下，结果大致与 Claude Fable 5 相当但成本更低，因此在您的评估显示质量保持不变的地方，降到 medium 或 low。在 low 下，Claude Fable 5.1 在每个任务的成本上通常与 Claude Opus 和 Claude Sonnet 模型具有竞争力，同时得分更高，因此在您原本会以较高 effort 级别运行较小模型的任何地方，都将其纳入比较。
有两个与 effort 相关的特定行为有各自的章节：在 low 下，Claude Fable 5.1 调用搜索和检索工具的频率更低（参见低 effort 下的搜索触发）；在 xhigh 和 max 下，它在撰写长篇交付物之前可能会思考更长时间（参见在 xhigh 和 max effort 下为长输出留出空间）。
与 Claude Fable 5 相比，Claude Fable 5.1 在长时间的工具调用轮次中可能会写出更少的面向用户的更新，尤其是在较高 effort 和较长的工具链中。用户会看到智能体一次沉默数分钟，或者最终消息只涵盖最后一步而非整个任务。
首先，检查您的客户端是否能收到进度更新。模型在工具调用之间的简短说明（它刚刚发现了什么以及接下来要做什么）会以进度更新 thinking 块的形式返回，而在默认的 thinking.display 为 "omitted" 时，这些块是空的。设置 display: "updates"（beta，需要 thinking-display-updates-2026-08-18 请求头）并将每个非空的 thinking 块渲染为一行状态信息，或者设置为 "summarized" 以便连同摘要化的推理一起接收它们。如果您没有请求它们，模型的更新可能根本没有到达您的用户。
其次，审查您的提示中是否有抑制叙述的指令。一些早期模型在工作时热衷于提供更新，这导致了诸如"将所有发现保留到最终回复中"之类的系统提示语句。在添加任何内容之前，先删除这类语句。
如果您仍然希望获得更多更新，例如在结对编程或其他"human-in-the-loop"（人在回路）工作中，请添加一行简短的系统提示，说明您何时希望模型输出面向用户的文本，以及每次更新应包含什么内容：
Before you start, say in a line what you're about to do; brief updates while you work help the user follow along. Close with a short recap that stands on its own — what you found, what you did, and what's next — so a reader who only sees the last message has the full picture.如果您的产品会折叠或隐藏工具输出，请告诉模型。否则它可能会运行命令来向用户"展示"您的 UI 从不显示的输出。请通过轮次范围的系统消息（clear_at: "next_user_message"，beta）传递该说明：
Only you see that command's output — the user's terminal shows at most a few lines of it. If the user needs to read any of it, put it in your reply.Claude Fable 5.1 通常会按预期发出并行工具调用：当请求指明了要获取的多个内容时，它会并行发出这些调用。例外情况是编码和计算机使用循环，其中下一批独立调用是由任务隐含而非明确请求的（自定义编码智能体、bash 加编辑器的 harness、计算机使用）：在这些场景中，它可能会每轮只发出一个调用。这不影响回答质量，但每多一轮都会消耗令牌、一次往返以及实际耗时。在当前请求末尾加一句简短的提示即可解决：
First privately list what you need next; then request every item that doesn't depend on another's result in this one response.每次您发回工具结果时，将其作为轮次范围的系统消息追加在该用户消息之后：即 messages 中一个带有 clear_at: "next_user_message" 的 role: "system" 条目。一旦存在更晚的用户消息，API 就会清除较早的副本，因此模型只会读取最新的那一条。轮次范围的系统消息处于 beta 阶段，需要 beta 请求头 mid-conversation-system-clear-at-2026-08-21。如果不使用该 beta，请改为将这句话放在同一用户消息中 tool_result 块之后的一个文本块里。
每轮追加一份新的副本，并将较早的副本原封不动地保留在原处，逐字节一致。它们仍留在数组中，但一旦被清除，模型就看不到它们，也不会消耗输入令牌。删除或重写它们属于对较早轮次的编辑：这会从该点起重新开始提示缓存，并使其后的 thinking 块失效（参见保持对话历史仅追加）。
以下循环展示了这种放置方式。每个 assistant 轮次都按返回时的原样发回，每个 user 轮次只携带工具结果，其后跟随一份新的轮次范围的提示副本。
import anthropic
from anthropic.types.beta import (
BetaMessageParam,
BetaToolParam,
BetaToolResultBlockParam,
)
client = anthropic.Anthropic()
BATCH_NUDGE = (
"First privately list what you need next; then request every item "
"that doesn't depend on another's result in this one response."
)
# 内存中的文件充当工作目录，使示例可在任何地方运行。
FILES = {
"pyproject.toml": """\
[project]
name = "demo"
version = "0.1.0"
description = "Demo project for the batching example"
""",
"README.md": """\
# demo
A small demo project. Run `demo --help` for usage.
""",
}
tools: list[BetaToolParam] = [
{
"name": "read_file",
"description": "Read a UTF-8 text file from the working directory.",
"input_schema": {
"type": "object",
"properties": {"path": {"type": "string"}},
"required": ["path"],
},
}
]
messages: list[BetaMessageParam] = [
{"role": "user", "content": "Summarize pyproject.toml and README.md."}
]
while True:
response = client.beta.messages.create(
model="claude-fable-5-1",
max_tokens=16000,
betas=["mid-conversation-system-clear-at-2026-08-21"],
tools=tools,
messages=messages,
)
# 按返回的原样追加 assistant 轮次，包括思考块。
messages.append({"role": "assistant", "content": response.content})
if response.stop_reason != "tool_use":
break
tool_results: list[BetaToolResultBlockParam] = []
for block in response.content:
if block.type == "tool_use":
path = str(block.input["path"])
if path in FILES:
tool_results.append(
{
"type": "tool_result",
"tool_use_id": block.id,
"content": FILES[path],
}
)
else:
tool_results.append(
{
"type": "tool_result",
"tool_use_id": block.id,
"content": f"File not found: {path}",
"is_error": True,
}
)
# 将工具结果作为 user 轮次发送，然后将提示的新副本作为
# 轮次范围的系统消息发送。保留先前的副本：API 会清除它们，
# 因此模型只会看到最新的一个。
messages.append({"role": "user", "content": tool_results})
messages.append(
{"role": "system", "content": BATCH_NUDGE, "clear_at": "next_user_message"}
)
print(next((block.text for block in response.content if block.type == "text"), ""))将每个 assistant 轮次按 API 返回的原样追加到历史中（包括 thinking 块），并且不要在请求之间编辑较早的轮次。对于在 2026 年 8 月 31 日或之后创建的新账户，Claude Fable 5.1 的 thinking 块仅在产生它们的那个确切对话中有效：如果某个请求在其前缀（系统提示、工具列表或任何较早的消息）发生变化后重放 thinking 块，将返回 400；或者如果您设置了 thinking.block_binding.prefix_mismatch_behavior: "drop_block"（beta，需要 thinking-binding-controls-2026-08-01 请求头），则会丢弃受影响的块。预计未来的模型将对所有账户强制执行此检查，因此即使您的账户目前未被强制执行，也请现在就采用这种模式。
会触发该检查的历史编辑，正是那些会重新开始提示缓存的编辑：注入和移除每轮提醒、就地摘要较早的轮次，或在会话中途更改系统提示。请将每轮提醒作为轮次范围的系统消息发送，使用对话中途系统消息来更改指令或工具，而不是重写 system 或 tools，并让服务端压缩或上下文编辑来完成任何裁剪。如果您在客户端进行压缩，最简单的形式是用一条摘要消息加上新的用户轮次替换整个历史，其他内容一概不重放：没有 thinking 块被带过来，因此不会有任何失败，模型会在压缩后的对话上重新思考（参见客户端自定义压缩）。
要找出您的 harness 已经在进行的编辑，请按照如何判断您的集成是否受影响中的描述，使用 prefix_mismatch_behavior: "drop_block" 运行一次会话并记录 input_transformations；或者捕获它在几个正常轮次中发送的确切请求，并确认连续请求在追加的轮次之前逐字节相同。
Claude Fable 5.1 的写作总体上比早期 Claude 模型更进一步，套话更少，未加解释的术语也更少。不过在某些情况下，它的行文比 Claude Fable 5 更密集：句子更长，段落分隔更少。一条定义了反模式（即矫饰文风）的指令会有帮助。将其添加到用户消息（首选）或系统提示中：
Mannered prose substitutes metaphor and flourish for direct statement. Instead of "a parameter worth varying," the mannered writer produces "a dial worth turning." Instead of "this point still matters," they write "this point earns its keep." The phrases exist to display the writer, not to convey the idea, and readers can tell. That is why mannered prose irritates: it makes the reader work harder so the writer can perform. It is also imprecise. Metaphors drag in connotations the writer did not choose and cannot control. The fix is to say what you mean. When a literal phrase is available, use it.简短版本通常也有效：
Please remove all mannered prose.早期模型在聊天中过度使用项目符号和粗体，许多提示中带有为抑制这一点而编写的反格式化规则。Claude Fable 5.1 则倾向于另一个方向：它更少使用粗体，也更不倾向于使用标题、列表或引号。如果您的提示包含反格式化的表述，请将其删除，或替换为一条说明何时适合使用特定格式的规则，例如：
Use lists and bullet points when asked to, or when the content is multifaceted enough that they help with clarity. If the person explicitly requests minimal formatting, always format your responses without bullet points, headers, lists, or bold emphasis, as requested. In conversational, personal, or emotional exchanges, keep to plain prose.在总结文档时，Claude Fable 5.1 比 Claude Fable 5 更有可能复述来源文本的段落而不将其标注为引用。要解决这个问题，请在系统提示中添加一个完整的正确回复示例：用户的请求、回复，以及一句解释该回复为何正确的话。
<example>
<user>look up how the Riverton Ledger and the Coast Dispatch each covered the Harbor Bridge closure and compare their reporting</user>
<response>
[web_search: Harbor Bridge closure Riverton Ledger]
[web_search: Harbor Bridge closure Coast Dispatch]
Both outlets agree on the basics: the bridge closed on March 3 after inspectors found cracked welds, and the state expects repairs to take about eight months. Where they differ is emphasis. The Ledger treats it as a local-economy story. The Dispatch frames it as a funding failure; its editorial calls the closure "entirely foreseeable." Read together, the Ledger explains who is affected now and the Dispatch explains how it came to this — neither account alone gives the whole picture.
</response>
<rationale>CORRECT: The response is organized around where the two outlets agree and differ, not as a walk through either article. Each outlet's reporting is conveyed in one or two sentences of the assistant's own indirect speech. One short marked phrase from one source; every other claim is reworded. The response is still specific and complete.</rationale>
</example>将两行 [web_search: ...] 替换为您自己工具的名称，以便模型将它们理解为模板化的工具输出，而不是要原样输出的字面文本。
Claude Fable 5.1 可以在没有太多方法论指导的情况下执行非常长的任务，尤其是在目标明确时。不过，在复杂的异步工作负载上，请提示它不要在工作完成之前结束轮次。如果没有这一提示，模型有时会描述它接下来要做什么而不是去做（"接下来，我将……"），或者停下来为原始请求已经涵盖的步骤征求许可（"我要应用这个吗？"）。用户不得不回复"继续"或"去做吧"，这适合结对编程和其他人在回路的工作，但没有发挥模型完整的长程能力。
两处系统提示补充共同缓解了这一问题。请同时应用两者。如果您需要限制提示长度，只使用第一处即可保留大部分效果。第一处告诉模型不要就已经请求的工作发问，并执行它已声明的后续步骤：
You are operating autonomously. The user is not watching in real time and cannot answer questions mid-task, so asking 'Want me to…?' or 'Shall I…?' will block the work. For reversible actions that follow from the original request, proceed without asking. Stop only for destructive actions or genuine scope changes the user must decide. Offering follow-ups after the task is done is fine; asking permission before doing the work is not.
Exception: when the user is describing a problem, asking a question, or thinking out loud rather than requesting a change, the deliverable is your assessment. Report your findings and stop. Don't apply a fix until they ask for one.
Before ending your turn, check your last paragraph. If it is a plan, an analysis, a question, a list of next steps, or a promise about work you have not done ('I'll…', 'let me know when…'), do that work now with tool calls. That includes retrying after errors and gathering missing information yourself. Do not stop because the context or session is long. End your turn only when the task is complete or you are blocked on input only the user can provide.
Before running a command that changes system state (such as restarts, deletes, or config edits), check that the evidence actually supports that specific action. A signal that pattern-matches to a known failure may have a different cause.开头那句告诉模型用户没有在观看的话承载了大部分效果。请保持原样。如果您的产品需要模型为特定确认而停下，请在其后添加一句话列出这些确认。此段落也可能使模型更不倾向于就含糊的请求发问，因此请在您自己的任务上检查这一权衡。
第二处将用户的请求定义为交付物的范围：
# Delivering work
The user's request — or the plan they approved — sets the scope, and the scope is the deliverable: don't quietly narrow, widen, or swap it. Read ambiguity the way a careful colleague would: make routine judgment calls yourself, and check in only when different readings would lead to materially different work. If you see a real problem with the task as specified, say so in a sentence or two and keep building under stated assumptions; if the user hears the concern and reaffirms, that is their decision, so deliver the full request.
If a question comes up partway, first do everything that doesn't depend on the answer; then state the assumption you made, or — when going ahead on a wrong guess would be unsafe or would make the work useless — put the question at the end of a turn that also delivers that progress. If one part turns out to be blocked, complete every other part in full and say exactly what you left out and why — the whole task is the deliverable, and scaling it down is the user's call, not yours. A step you have decided on is something to run, not to announce: describing the next step and ending the turn leaves it undone until the user replies.
Keep changes to what the request needs. Something else you notice worth doing — cleanup or documentation the task didn't call for, a change to a file the task didn't require — is a suggestion to make at the end, not a change to make; actions clearly beyond what the ask implies, and risky or destructive ones, still need the user's go-ahead.当长对话被压缩时，明确告知 Claude Fable 5.1 其摘要必须保留什么内容，效果很好。服务端压缩已经做到了这一点。如果您在客户端进行压缩，请使用以下摘要指令：
Summarize the transcript inside <summary></summary> tags. Include relevant information in the summary such that this conversation will be continued by a new context window without needing to redo work or be reprovided with relevant constraints or context. Be sure to preserve: (1) any difficulties or problems that came up, and how they were handled or resolved; (2) any possibilities, options, or approaches that were raised, tried, or set aside, and why; (3) anything that was asked for, decided, agreed, ruled out, or established as a preference, constraint, or boundary — stated exactly; (4) exactly where things stand now — what has been covered, settled, or completed so far; (5) anything still open, unresolved, promised, or expected to happen next; (6) specific details that would be hard to reconstruct — names, numbers, dates, exact wording, links or references — kept exactly. Be complete on these even at the cost of length; keep everything else concise. Weight the two voices differently: keep what the user said, asked for, shared, or established carefully and close to their own words; your own explanations and reasoning can be condensed much further, to what they concluded or produced — as long as nothing in the six items above is dropped.当被要求实现一个开放式功能时，Claude Fable 5.1 会交付所要求的内容，有时还会更多：它可能会修复附近的代码、扩展任务未提及的行为，或提交多于该更改所需的测试文件。它对关于应省略什么的明确指令响应良好。使用以下指令后，未经请求的添加和提交的测试代码大幅减少，而任务成功率没有可测量的变化：
If, while working or testing, you find a pre-existing bug, a performance concern, or behavior the task doesn't mention, don't fix, optimize or extend it in this change unless the requested behavior cannot work without it; report it as a follow-up in your summary. Where the task is ambiguous, implement the reading its wording and the surrounding code most directly support, state that assumption in your summary, and don't build for the other readings as well. Verify your work however you like; scratch scripts and quick checks need not be kept. Commit tests only where the task asks for them or this repository already keeps tests for this kind of change, sized like the neighboring test files — roughly one focused test per stated behavior — and don't turn scratch checks into additional permanent test files. This is about extras only: implement every behavior the task asks for, completely.在 low effort 下，与 Claude Fable 5 相比，Claude Fable 5.1 不太可能调用搜索或检索工具，而更可能从记忆中回答。在某些情况下，最简单的解决方法是为受影响的轮次而不是整个对话提高 effort。请参阅在对话中途更改 effort。
在其他情况下，一条引导其进行验证的提示会有帮助。在系统提示中说明，认出一个名称并不等于了解其当前状态，并且此类名称应按用户所写的原样进行搜索：
When a query centers on a name you do not confidently recognize, or recognize from a fast-moving area like AI models and developer tools where the landscape shifts within months, the name itself is the thing to verify: search before answering, and include the name as the user wrote it in at least one query alongside any reformulations. This holds even when you have some background on it — partial background is exactly what makes an out-of-date answer sound authoritative, so familiarity is not a reason to skip the search.Claude Fable 5.1 的安全分类器产生的误报比 Claude Fable 5 发布时更少，并且允许在源代码中查找漏洞。误报仍会发生，被拦截的请求会返回 stop_reason: "refusal"（参见拒绝、回退与计费）。有三种情况会使误报更容易发生：
如果 Claude Fable 5.1 为小改动重写整个文件，请将以下指令追加到系统提示或第一条用户消息中。Claude Fable 5.1 比 Claude Fable 5 更有可能重写整个文本文件而不是进行定向编辑。生成的文件通常是相同的，但除非文件很短或大部分内容都在变化，否则重写会消耗更多的输出令牌和时间。该指令使 Claude Fable 5.1 在小型和中型更改上回到与 Claude Fable 5 一致的行为。
The number of tokens used to edit files is best minimized, all else being equal. Therefore, when it will not affect the end result, try to surgically edit a file rather than rewrite the entire thing.在 xhigh 尤其是 max effort 下，Claude Fable 5.1 在开始撰写回复之前可能会思考更长时间。当单个请求要求一份长篇交付物（例如对一份长文档的完整重写）时，它可能会在思考中起草该交付物的大部分内容，然后再将其作为回复重新写出，这意味着更长的等待和更多的输出令牌。最简单的做法是在 high（推荐的起点）下运行此类请求，仅在您测量到质量提升的地方才转到 xhigh 或 max（参见考虑所有 effort 级别）。如果您确实在 xhigh 或 max 下运行它们：
max_tokens 时要为思考和回复都留出空间，而不仅仅是您预期的回复长度。[max_tokens] 替换为该请求实际的 max_tokens 值，例如 64,000。Everything Claude produces in one reply, including any reasoning or drafting it does before the reply, counts toward a single limit of about [max_tokens] tokens. If that limit is reached before the reply is finished, the person receives a cut-off response and has to start over. Composing an entire output or deliverable in full as reasoning and then again as a reply would double the length of the turn without improving the result, so Claude doesn't do that.
Instead, when the person has asked for a long or effort-intensive deliverable such as a multi-section document, a large table or dataset, or a complete code file, Claude spends extra effort on understanding the request, checking the inputs Claude's answer depends on, settling the structure and other difficult decisions, and otherwise using the reasoning space to reason and the output space to write an output. If Claude plans well then it should not need to draft its output multiple times (and Claude is pretty good at planning, so this should not be an issue).如果您的编码智能体允许 Claude Fable 5.1 将工作委派给子智能体，请不要强制主智能体停下来等待每一个子智能体。在编码任务上，让主智能体在子智能体运行时继续工作，可以在质量、令牌用量和成本相近的情况下降低平均完成时间。要进行此设置：
user 消息中将其传回给主智能体。模型仍然经常选择等待。时间上的节省来自它继续进行其他工作的那些运行。
Claude Fable 5.1 开箱即具备更好的视觉能力，在密集图表等复杂视觉输入上，当它能够迭代地分析、裁剪并以视觉方式验证所见内容时，表现最佳。要获得全部收益，请将模型作为智能体运行，使其能够访问一个存放原始图像或视频、并预装了基本图像处理库（如 PIL 和 OpenCV）的容器。如果运行容器的开销过大，仅一个图像裁剪工具就能带来大部分提升：一个返回图像中所选区域（经裁剪并放大）的工具，可以让模型更深入地检查特定细节，并使测试时计算量随图像令牌扩展。裁剪工具示例中有一个可用的定义。
Was this page helpful?