Claude 3 重磅发布！Anthropic 发布了新一代AI大模型系列

type

status

date

slug

summary

Claude 3 model family Claude 3 模型系列

A new standard for intelligence

智能新标准

Opus, our most intelligent model, outperforms its peers on most of the common evaluation benchmarks for AI systems, including undergraduate level expert knowledge (MMLU), graduate level expert reasoning (GPQA), basic mathematics (GSM8K), and more. It exhibits near-human levels of comprehension and fluency on complex tasks, leading the frontier of general intelligence.

Opus 是我们最智能的模型，在大多数常见的 AI 系统评估基准上都优于同行，包括本科水平专家知识（MMLU）、研究生水平专家推理（GPQA）、基础数学（GSM8K）等。它对复杂任务表现出接近人类的理解力和流利度，引领着一般智能的前沿。

All Claude 3 models show increased capabilities in analysis and forecasting, nuanced content creation, code generation, and conversing in non-English languages like Spanish, Japanese, and French.

所有 Claude 3 模型在分析和预测、细致入微的内容创建、代码生成以及使用西班牙语、日语和法语等非英语语言进行交谈方面都显示出增强的能力。

Below is a comparison of the Claude 3 models to those of our peers on multiple benchmarks [1] of capability:以下是 Claude 3 型号与我们的同行在多个性能基准 [1] 上的比较：

Near-instant results

近乎即时的结果

The Claude 3 models can power live customer chats, auto-completions, and data extraction tasks where responses must be immediate and in real-time.

Claude 3 型号可以支持实时客户聊天、自动完成和数据提取任务，在这些任务中，响应必须是即时和实时的。

Haiku is the fastest and most cost-effective model on the market for its intelligence category. It can read an information and data dense research paper on arXiv (~10k tokens) with charts and graphs in less than three seconds. Following launch, we expect to improve performance even further.

Haiku 是市场上最快、最具成本效益的智能类别模型。它可以在不到三秒的时间内读取有关arXiv（~10k个代币）的信息和数据密集的研究论文，其中包含图表和图形。发布后，我们希望进一步提高性能。

For the vast majority of workloads, Sonnet is 2x faster than Claude 2 and Claude 2.1 with higher levels of intelligence. It excels at tasks demanding rapid responses, like knowledge retrieval or sales automation. Opus delivers similar speeds to Claude 2 and 2.1, but with much higher levels of intelligence.

对于绝大多数工作负载，Sonnet 比 Claude 2 和 Claude 2.1 快 2 倍，具有更高的智能水平。它擅长需要快速响应的任务，例如知识检索或销售自动化。Opus 提供与 Claude 2 和 2.1 相似的速度，但智能水平要高得多。

Strong vision capabilities

强大的视觉能力

The Claude 3 models have sophisticated vision capabilities on par with other leading models. They can process a wide range of visual formats, including photos, charts, graphs and technical diagrams. We’re particularly excited to provide this new modality to our enterprise customers, some of whom have up to 50% of their knowledge bases encoded in various formats such as PDFs, flowcharts, or presentation slides.

Claude 3 型号具有与其他领先型号相媲美的复杂视觉功能。它们可以处理各种视觉格式，包括照片、图表、图形和技术图表。我们特别高兴能为我们的企业客户提供这种新模式，其中一些客户拥有多达 50% 的知识库以各种格式编码，例如 PDF、流程图或演示幻灯片。

Fewer refusals

减少拒绝

Previous Claude models often made unnecessary refusals that suggested a lack of contextual understanding. We’ve made meaningful progress in this area: Opus, Sonnet, and Haiku are significantly less likely to refuse to answer prompts that border on the system’s guardrails than previous generations of models. As shown below, the Claude 3 models show a more nuanced understanding of requests, recognize real harm, and refuse to answer harmless prompts much less often.

以前的克劳德模型经常做出不必要的拒绝，这表明缺乏对上下文的理解。我们在这一领域取得了有意义的进展：与前几代模型相比，Opus、Sonnet 和 Haiku 拒绝回答系统护栏提示的可能性要小得多。如下图所示，Claude 3 模型对请求表现出更细致的理解，识别真正的伤害，并且拒绝回答无害提示的频率要低得多。

Improved accuracy

提高准确性

Businesses of all sizes rely on our models to serve their customers, making it imperative for our model outputs to maintain high accuracy at scale. To assess this, we use a large set of complex, factual questions that target known weaknesses in current models. We categorize the responses into correct answers, incorrect answers (or hallucinations), and admissions of uncertainty, where the model says it doesn’t know the answer instead of providing incorrect information. Compared to Claude 2.1, Opus demonstrates a twofold improvement in accuracy (or correct answers) on these challenging open-ended questions while also exhibiting reduced levels of incorrect answers.

各种规模的企业都依赖我们的模型来为他们的客户提供服务，因此我们的模型输出必须保持大规模的高精度。为了评估这一点，我们使用了大量复杂的事实问题，这些问题针对当前模型中的已知弱点。我们将回答分为正确答案、错误答案（或幻觉）和承认不确定性，其中模型说它不知道答案，而不是提供不正确的信息。与 Claude 2.1 相比，Opus 在这些具有挑战性的开放式问题的准确性（或正确答案）方面提高了两倍，同时也减少了错误答案的水平。

In addition to producing more trustworthy responses, we will soon enable citations in our Claude 3 models so they can point to precise sentences in reference material to verify their answers.

除了产生更可信的回答外，我们还将很快在我们的 Claude 3 模型中启用引用，以便他们可以指向参考资料中的精确句子来验证他们的答案。

Long context and near-perfect recall

长上下文和近乎完美的回忆

The Claude 3 family of models will initially offer a 200K context window upon launch. However, all three models are capable of accepting inputs exceeding 1 million tokens and we may make this available to select customers who need enhanced processing power.

Claude 3 系列型号最初将在发布时提供 200K 上下文窗口。但是，所有三种型号都能够接受超过 100 万个代币的输入，我们可能会将其提供给需要增强处理能力的特定客户。

To process long context prompts effectively, models require robust recall capabilities. The 'Needle In A Haystack' (NIAH) evaluation measures a model's ability to accurately recall information from a vast corpus of data. We enhanced the robustness of this benchmark by using one of 30 random needle/question pairs per prompt and testing on a diverse crowdsourced corpus of documents. Claude 3 Opus not only achieved near-perfect recall, surpassing 99% accuracy, but in some cases, it even identified the limitations of the evaluation itself by recognizing that the "needle" sentence appeared to be artificially inserted into the original text by a human.

为了有效地处理长上下文提示，模型需要强大的召回功能。“大海捞针”（NIAH）评估衡量模型从大量数据语料库中准确调用信息的能力。我们通过对每个提示使用 30 个随机针/问题对之一，并在不同的众包文档语料库上进行测试，增强了该基准的稳健性。Claude 3 Opus 不仅实现了近乎完美的回忆，准确率超过 99%，而且在某些情况下，它甚至通过识别“针”句似乎是人类人为地插入原始文本来识别评估本身的局限性。

Responsible design

负责任的设计

We’ve developed the Claude 3 family of models to be as trustworthy as they are capable. We have several dedicated teams that track and mitigate a broad spectrum of risks, ranging from misinformation and CSAM to biological misuse, election interference, and autonomous replication skills. We continue to develop methods such as Constitutional AI that improve the safety and transparency of our models, and have tuned our models to mitigate against privacy issues that could be raised by new modalities.

我们开发了 Claude 3 系列型号，使其既值得信赖又强大。我们有几个专门的团队来跟踪和缓解广泛的风险，从错误信息和 CSAM 到生物滥用、选举干扰和自主复制技能。我们继续开发诸如宪法人工智能之类的方法，以提高我们模型的安全性和透明度，并调整了我们的模型，以减轻新模式可能引发的隐私问题。

Addressing biases in increasingly sophisticated models is an ongoing effort and we’ve made strides with this new release. As shown in the model card, Claude 3 shows less biases than our previous models according to the Bias Benchmark for Question Answering (BBQ). We remain committed to advancing techniques that reduce biases and promote greater neutrality in our models, ensuring they are not skewed towards any particular partisan stance.

解决日益复杂的模型中的偏差是一项持续的努力，我们在这个新版本中取得了长足的进步。如模型卡所示，根据问答偏差基准（BBQ），Claude 3 比我们以前的模型显示出更少的偏差。我们仍然致力于推进技术，以减少偏见并促进我们模型中的更大中立性，确保它们不会偏向任何特定的党派立场。

While the Claude 3 model family has advanced on key measures of biological knowledge, cyber-related knowledge, and autonomy compared to previous models, it remains at AI Safety Level 2 (ASL-2) per our Responsible Scaling Policy. Our red teaming evaluations (performed in line with our White House commitments and the 2023 US Executive Order) have concluded that the models present negligible potential for catastrophic risk at this time. We will continue to carefully monitor future models to assess their proximity to the ASL-3 threshold. Further safety details are available in the Claude 3 model card.

虽然与以前的型号相比，Claude 3 型号系列在生物知识、网络相关知识和自主性等关键指标上取得了进步，但根据我们的负责任扩展政策，它仍处于 AI 安全级别 2 （ASL-2）。我们的红队评估（根据我们的白宫承诺和 2023 年美国行政命令进行）得出的结论是，这些模型目前存在灾难性风险的可能性可以忽略不计。我们将继续仔细监测未来的模型，以评估它们是否接近 ASL-3 阈值。Claude 3 车型卡中提供了更多安全细节。

Easier to use

更易于使用

The Claude 3 models are better at following complex, multi-step instructions. They are particularly adept at adhering to brand voice and response guidelines, and developing customer-facing experiences our users can trust. In addition, the Claude 3 models are better at producing popular structured output in formats like JSON—making it simpler to instruct Claude for use cases like natural language classification and sentiment analysis.

Claude 3 型号更擅长遵循复杂的多步骤说明。他们特别擅长遵守品牌声音和响应准则，并开发用户可以信赖的面向客户的体验。此外，Claude 3 模型更擅长以 JSON 等格式生成流行的结构化输出，从而可以更轻松地指导 Claude 进行自然语言分类和情感分析等用例。

Model details

型号详细信息

Claude 3 Opus is our most intelligent model, with best-in-market performance on highly complex tasks. It can navigate open-ended prompts and sight-unseen scenarios with remarkable fluency and human-like understanding. Opus shows us the outer limits of what’s possible with generative AI.

Claude 3 Opus 是我们最智能的型号，在高度复杂的任务上具有市场上最好的性能。它可以以非凡的流畅性和类似人类的理解来导航开放式提示和看不见的场景。Opus 向我们展示了生成式 AI 可能性的外部极限。

Cost 成本[Input $/million tokens \| Output $/million tokens][输入$/百万代币 \|输出 $/million 代币]	$15 \| $75 $15 \|75美元
Context window 上下文窗口	200K* 200千米*
Potential uses 潜在用途	• Task automation: plan and execute complex actions across APIs and databases, interactive coding任务自动化：跨 API 和数据库规划和执行复杂操作，交互式编码 • R&D: research review, brainstorming and hypothesis generation, drug discovery研发：研究回顾、头脑风暴和假设生成、药物发现 • Strategy: advanced analysis of charts & graphs, financials and market trends, forecasting策略：对图表和图形、财务和市场趋势的高级分析、预测
Differentiator 微分	Higher intelligence than any other model available.比任何其他可用型号都具有更高的智能性。

1M tokens available for specific use cases, please inquire.

Claude 3 Sonnet strikes the ideal balance between intelligence and speed—particularly for enterprise workloads. It delivers strong performance at a lower cost compared to its peers, and is engineered for high endurance in large-scale AI deployments.

克劳德 3 十四行诗在智能和速度之间取得了理想的平衡，尤其是对于企业工作负载。与同类产品相比，它以更低的成本提供强大的性能，专为大规模 AI 部署而设计，具有高耐用性。

Cost 成本[Input $/million tokens \| Output $/million tokens][输入$/百万代币 \|输出 $/million 代币]	$3 \| $15 $3 \|15美元
Context window 上下文窗口	200K 200千米赛
Potential uses 潜在用途	• Data processing: RAG or search & retrieval over vast amounts of knowledge数据处理：RAG或搜索和检索大量知识 • Sales: product recommendations, forecasting, targeted marketing销售：产品推荐、预测、有针对性的营销 • Time-saving tasks: code generation, quality control, parse text from images节省时间的任务：代码生成、质量控制、从图像中解析文本
Differentiator 微分	More affordable than other models with similar intelligence; better for scale.比其他具有类似智能的型号更实惠;更适合规模。

Claude 3 Haiku is our fastest, most compact model for near-instant responsiveness. It answers simple queries and requests with unmatched speed. Users will be able to build seamless AI experiences that mimic human interactions.

Claude 3 Haiku 是我们速度最快、最紧凑的型号，具有近乎即时的响应能力。它以无与伦比的速度回答简单的查询和请求。用户将能够构建模仿人类交互的无缝 AI 体验。

Cost 成本[Input $/million tokens \| Output $/million tokens][输入$/百万代币 \|输出 $/million 代币]	$0.25 \| $1.25 0.25美元 \|1.25美元
Context window 上下文窗口	200K 200千米赛
Potential uses 潜在用途	• Customer interactions: quick and accurate support in live interactions, translations客户互动：在现场互动、翻译方面提供快速准确的支持 • Content moderation: catch risky behavior or customer requests内容审核：捕获有风险的行为或客户请求 • Cost-saving tasks: optimized logistics, inventory management, extract knowledge from unstructured data节省成本的任务：优化物流、库存管理、从非结构化数据中提取知识
Differentiator 微分	Smarter, faster, and more affordable than other models in its intelligence category.比其智能类别中的其他型号更智能、更快、更实惠。

Model availability

型号可用性

Opus and Sonnet are available to use today in our API, which is now generally available, enabling developers to sign up and start using these models immediately. Haiku will be available soon. Sonnet is powering the free experience on claude.ai, with Opus available for Claude Pro subscribers.

Opus 和 Sonnet 现已在我们的 API 中可用，该 API 现已正式发布，使开发人员能够立即注册并开始使用这些模型。俳句即将推出。Sonnet 正在为 claude.ai 上的免费体验提供支持，Opus 可供 Claude Pro 订阅者使用。

Sonnet is also available today through Amazon Bedrock and in private preview on Google Cloud’s Vertex AI Model Garden—with Opus and Haiku coming soon to both.

十四行诗今天也可通过 Amazon Bedrock 获得，并在 Google Cloud 的 Vertex AI Model Garden 上提供私人预览版——Opus 和 Haiku 即将推出。

Smarter, faster, safer

更智能、更快、更安全

We do not believe that model intelligence is anywhere near its limits, and we plan to release frequent updates to the Claude 3 model family over the next few months. We're also excited to release a series of features to enhance our models' capabilities, particularly for enterprise use cases and large-scale deployments. These new features will include Tool Use (aka function calling), interactive coding (aka REPL), and more advanced agentic capabilities.

我们认为模型智能不会接近其极限，我们计划在未来几个月内频繁发布 Claude 3 模型系列的更新。我们也很高兴发布一系列功能来增强我们模型的功能，特别是对于企业用例和大规模部署。这些新功能将包括工具使用（又名函数调用）、交互式编码（又名 REPL）和更高级的代理功能。

As we push the boundaries of AI capabilities, we’re equally committed to ensuring that our safety guardrails keep apace with these leaps in performance. Our hypothesis is that being at the frontier of AI development is the most effective way to steer its trajectory towards positive societal outcomes.

随着我们不断突破 AI 功能的界限，我们同样致力于确保我们的安全护栏跟上这些性能飞跃的步伐。我们的假设是，处于人工智能发展的前沿是引导其走向积极社会成果的最有效方式。

We’re excited to see what you create with Claude 3 and hope you will give us feedback to make Claude an even more useful assistant and creative companion. To start building with Claude, visit anthropic.com/claude.

我们很高兴看到您使用 Claude 3 创作的内容，并希望您能向我们提供反馈，使 Claude 成为更有用的助手和创意伴侣。要开始与 Claude 一起构建，请访问 anthropic.com/claude。

Footnotes

脚注 1. This table shows comparisons to models currently available commercially that have released evals. Our model card shows comparisons to models that have been announced but not yet released, such as Gemini 1.5 Pro. In addition, we’d like to note that engineers have worked to optimize prompts and few-shot samples for evaluations and reported higher scores for a newer GPT-4T model. Source.

下表显示了与目前市面上已发布 eval 的模型的比较。我们的模型卡显示了与已宣布但尚未发布的模型的比较，例如 Gemini 1.5 Pro。此外，我们想指出的是，工程师们一直在努力优化用于评估的提示和小样本，并报告了较新的 GPT-4T 模型的更高分数。源。