嗨 Claude！你能把一个复杂的主题讲得简单明了吗？如果你需要我提供更多信息，请立即问我1-2个关键问题。如果你觉得我应该上传任何有助于你更好完成任务的文件，请告诉我。你可以使用你可以访问的工具，比如 Google Drive、网络搜索等——如果它们能帮助你更好地完成这个任务的话。不要使用分析工具。请保持你的回答友好、简短和对话式。请尽快执行任务——如果有一个可行的产物会很好。如果使用产物，请考虑哪种类型的产物（互动的、视觉的、清单等）对这个具体任务最有帮助。谢谢你的帮助！
嗨 Claude！你能帮我理清这些想法吗？如果你需要我提供更多信息，请立即问我1-2个关键问题。如果你觉得我应该上传任何有助于你更好完成任务的文件，请告诉我。你可以使用你可以访问的工具，比如 Google Drive、网络搜索等——如果它们能帮助你更好地完成这个任务的话。不要使用分析工具。请保持你的回答友好、简短和对话式。请尽快执行任务——如果有一个可行的产物会很好。如果使用产物，请考虑哪种类型的产物（互动的、视觉的、清单等）对这个具体任务最有帮助。谢谢你的帮助！
嗨 Claude！你能帮我准备考试或面试吗？如果你需要我提供更多信息，请立即问我1-2个关键问题。如果你觉得我应该上传任何有助于你更好完成任务的文件，请告诉我。你可以使用你可以访问的工具，比如 Google Drive、网络搜索等——如果它们能帮助你更好地完成这个任务的话。不要使用分析工具。请保持你的回答友好、简短和对话式。请尽快执行任务——如果有一个可行的产物会很好。如果使用产物，请考虑哪种类型的产物（互动的、视觉的、清单等）对这个具体任务最有帮助。谢谢你的帮助！
One of the most frequent questions we hear is “what model should I choose for this workload?” As we have released more model classes and versions, the answer has become more nuanced.
This article covers those details including a description of each model class, the top questions to ask when selecting a model, and other best practices.
But to put aside the nuance for a moment, our default recommendation is to start with the most intelligent generally available model and use effort level to dial in performance and cost.
Cost-per-task is often lower for more intelligent models, especially at lower effort levels, even if the price-per-token is higher. This is because more capable models often take fewer turns and less thinking time to get most tasks right. Starting with a smaller model can also make it harder to distinguish between model failures and setup failures.
Of course, as use cases arise that are more latency or cost-sensitive, you can test lower tier models until you find your ideal fit.
Some organizations may also choose to start with the most cost effective model and move up classes until the quality bar is met. We include both directional approaches：https://platform.claude.com/docs/en/about-claude/models/choosing-a-model in our documentation on model selection.
Mythos is Anthropic’s most capable model class, with frontier capabilities across domains. This model class is especially capable at coding, long-running agent tasks, and solving problems AI has not reliably handled before.
Opus is our powerful model class for reasoning-intensive enterprise tasks. Opus models consistently rank among leading models on key industry benchmarks such as GDPval-AA for knowledge work and Terminal-Bench 2.1 for agentic coding.
The choice between Opus and Fable may not seem clear on the surface, as both excel at coding, long-running agents, and knowledge work. In real-world situations, larger models such as Fable tend to have more wisdom, creativity, and writing skills despite having similar benchmark scores to models such as Opus.
The general rule of thumb is if your evals or internal testing show Opus struggling on some tasks, then Fable is the answer. If Opus already clears the quality bar, then its speed and price profile may make it the better choice.
Sonnet is our versatile model class for everyday tasks. Sonnet provides a balance of performance, cost, and speed for the widest set of general purpose use cases, including high-volume sub-agents in multi-agent orchestration setups.
Haiku is our lowest cost and fastest model class. Haiku models are designed for high-frequency workloads where latency and cost matter.
Our model classes don’t specialize in one type of work. We don’t recommend one model class for finance and another for science. Every Claude model is trained to excel in areas like coding, agentic tasks, and knowledge work.
The main difference across model classes is in how hard a problem they can reliably carry, and what that capability costs in price and speed. When choosing a model, ask:
How hard is this task? If it typically takes a lot of time, involves multiple steps, or is previously unsolved then a more capable model class is appropriate.
What are the latency needs? If the model is involved in high-frequency customer facing workloads, then Sonnet is often the best choice.
What are the access constraints? Mythos is only available to organizations under Project Glasswing：https://www.anthropic.com/glasswing. Not all organizations make all model classes available to all roles.
What are the unit economics ? Higher volumes of production may be more appropriate for lower classes of models, particularly if evaluations show those tasks are completed satisfactorily. Models are priced differently per token：https://platform.claude.com/docs/en/about-claude/models/overview and will have different price-per-task costs based on their capabilities and effort level.
Effort level also impacts the balance of quality, speed, and cost. Higher-class models at higher efforts offer the best possible performance, and higher-class models at lower efforts can sometimes be more efficient than smaller models.
To learn more read Choosing a Claude model and effort level in Claude Code：https://claude.com/blog/claude-model-and-effort-level-in-claude-code.
The advisor strategy：https://claude.com/blog/the-advisor-strategy allows faster, lower-cost worker models to call more intelligent models to check their plan and evaluate their work, leading to improved performance.
This method, where the executor model is coached only when needed, improves performance by a substantial amount. For example, on SWE-bench Pro Sonnet 5 with a Fable 5 advisor is within 10% of Fable 5’s score at 63% of the price of using Fable 5 for the whole task.
Two common ways to see if model capabilities are sufficient for your needs are to use standard benchmarks and custom evaluations.
Benchmarks are a set of pre-determined tasks or scenarios, often for a specific domain, with known solutions. These can be helpful directional guides for evaluating capabilities across model classes and providers. The challenge arises when evaluating powerful models, such as Opus and Fable, which can solve almost all of the questions on the test (often referred to as saturation).
In these cases, we recommend organizations use the models on real workloads or test them with their own evaluations to make a decision on which model is the right choice. Typically, evaluations are a curated set of problems drawn from production — including difficult tasks where your current tools fall short, with success criteria your team defines.
This is where the capability and creativity of frontier models start to separate from the pack and from one another. We’ve written extensively on the best practices for developing custom agent evaluations：https://www.anthropic.com/engineering/demystifying-evals-for-ai-agents.
There is no one-size-fits-all approach to AI model selection, which is why we make multiple model classes available. Ultimately, the best way to select a model is to understand the basics of each model class and understand your use case in-depth. That means building, maintaining, and deploying strong evaluations.
Explore more product news and best practices for teams building with Claude.
Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.
Hi Claude! Could you help me develop a unique voice for an audience? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you improve my writing style? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you brainstorm creative ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you explain a complex topic simply? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you help me make sense of these ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you prepare for an exam or interview? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you explain a programming concept? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you look over my code and give me tips? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you vibe code with me? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
Hi Claude! Could you write grant proposals? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to — like Google Drive, web search, etc. — if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can - an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help!
情报判断
Aioga 编辑摘要
Aioga 编辑摘要：Anthropic 发布 Claude 模型选择指南，将模型分为 Mythos/Fable（最强能力，用于前沿编码与智能体任务）、Opus（推理密集型企业任务）、Sonnet（日常通用任务）和 Aioga 将其归入「技巧观点」方向，重点关注它对真实使用和行业竞争的影响。
扫码直达文章详情