Claude Blog 采集 (2026-07-25)¶
共采集 5 篇文章
📋 文章索引¶
- Four role-based certifications for the people who put Claude to work for customers - Jul 23, 2026 (评分: 9.5)
- Claude models explained: choosing the best model for your use case - Jul 24, 2026 (评分: 9.0)
- Working at the frontier: How Cursor knew Claude Fable 5 was ready for the hardest 1% of problems - Jul 17, 2026 (评分: 9.0)
- Working at the frontier: How Cognition trusts Claude Fable 5 to work through the night - Jul 10, 2026 (评分: 9.0)
- Think through hard problems in voice mode - Jul 23, 2026 (评分: 8.2)
Four role-based certifications for the people who put Claude to work for customers¶
来源: Claude Blog 发布日期: Jul 23, 2026 采集时间: 2026-07-25 价值评分: 9.5/10 正文字数: ~7898 字符
摘要¶
We're expanding the Claude Certification Program with three new role-based certifications that span the full team a customer needs to deploy AI.
正文内容¶
We're expanding the Claude Certification Program with three new role-based certifications that span the full team a customer needs to deploy AI. Anthropic's goal is to be the leader of AI enablement. In support of this goal, we launched the Claude Partner Network in March alongside a $100 million investment and a Services Track in June to help enterprise buyers find the right firms to put Claude to work inside their business. Part of helping them find the right firm is ensuring their people are properly certified—and we want Claude certifications to be a benchmark for quality that customers, partners, and the industry recognize as verified expertise, not just a badge. To achieve this, Claude certifications are granted through supervised, identity-verified exams that test the specific skills a role calls for. This approach is quickly gaining traction. Since launching in March, more than 36,000 consultants have received certification across more than 1,300 organizations. More than 400,000 people have completed Claude training this year through Anthropic Partner Academy, our free training platform for partners. Here’s what you need to know about our expanded certification program. The four credentials Putting Claude to use at a large enterprise involves different roles and specialties, and the certification maps to the four largest ones: (New) Claude Certified Associate: Foundations validates practical, everyday use of Claude for anyone working on Claude-related projects. This applies to a wide range of roles including consultants, project leads, and both business and technical expertise. (New) Claude Certified Developer: Foundations is for engineers building applications with Claude, and includes training on the Claude API, tool use, and agent development. (New) Claude Certified Architect: Professional is the advanced credential, covering integration architecture, governance, and evaluation, built for the scale of large enterprises. Claude Certified Architect: Foundations is for solution architects who design and build agent systems with Claude. Every path to getting credentialed starts with a foundation-level certification and advances to the professional-level. Offerings span non-technical roles through those that build and architect Claude-based solutions, and each certification was designed by subject matter experts so the content reflects the real work that newly certified professionals will do. Completed certifications are recognized with partner badges, which show demonstrated expertise in a specific product area. "The companies that pull ahead over the next twelve months will be the ones with enough skilled people to put Claude to work on their hardest problems. Every major technology shift has been carried by a trained, trusted workforce, and AI is no different. That's what we're building with our partners, and it's why the world's largest firms are certifying their people by the tens of thousands." – Steve Corfield, head of global business development and partnerships. Firms across the network are committing to certify their people at scale and are prioritizing certification during the hiring process, and the Global Premier partners are doing it in the largest cohorts. Accenture has committed to have 50,000 certified professionals across foundational, technical, and specialized tracks. Capgemini will certify 20,000 professionals. Cognizant will certify 10,000 developers and engineers. Deloitte will certify 15,000 professionals. DXC has committed to 20,000 certifications. Infosys has committed to 10,000. KPMG will certify 15,000 professionals. PwC will certify 30,000 professionals. TCS has committed to 10,000. And UST plans to certify 20,000 individuals worldwide, across roles from architects and engineers to consultants and industry specialists. It isn't only the largest firms: specialist AI-native and boutique consultancies are certifying entire delivery teams. Ascendion plans to certify its entire 8,000-person engineering team before 2027. EPAM will have more than 10,000 Claude-certified architects, including 250 specialized forward-deployed engineers. Fractal is targeting 2,000 certified practitioners. Globant is growing a practice toward 5,000 certified architects. Persistent expects more than 3,000 Claude certifications by the end of 2026. And Wipro will certify 10,000 front-line delivery experts within 18 months. What certification means for customers Our Claude Certification Program is designed to give enterprises confidence in the partner they are selecting. Certification represents validated capability, rather than course attendance. Every exam is proctored and identity-verified, and each was designed by subject matter experts, including Anthropic’s own Applied AI team, to reflect real work that’s done in the field A certification establishes that this individual has demonstrated the skills their role demands. Credentials are matched to role-specific skills. Each credential maps to the four roles that put Claude into production, from everyday practical use, to building with the Claude API, to architecting large-scale enterprise deployments. This is designed so the buyer knows what a certified person was tested on. The Claude Partner Network describes a firm’s demonstrated capacity. . Tier standing in the Claude Partner Network combines certified practitioners with deployed customers and public customer references. How the exams work We work with Pearson, drawing on its global assessment and credentialing infrastructure, to deliver our program. Every exam is proctored, meaning it is taken under supervision, and delivered through Pearson Professional Assessments. Exams are securely administered, and test takers must validate their identity before beginning an exam. Those who pass the exam receive a digital badge through Credly by Pearson. "AI is moving faster than most people's ability to build the skills to use it confidently, and that gap is exactly what verified assessment is built to close. Anthropic has built the first proctored certification among AI labs that uses the same trusted infrastructure for skills, credentials, and verified outcomes that Pearson uses across professional licensure. This is another example that shows why Anthropic is a trusted AI partner for businesses globally." – Vishaal Gupta, President, Enterprise Learning and Skills, Pearson Proctoring and identity verification matter because these credentials are used to make decisions. Certified practitioner counts determine a firm's standing in the Claude Partner Network and are relevant when companies vet who will work on their systems. A credential can only support those decisions if the assessment is rigorous and validates a person’s applied skills. Our certification is gaining momentum inside the consulting and engineering firms implementing Claude for their clients. Certifications also determine how firms advance in the network. The top tier, Global Premier, requires 1,000 certified practitioners, 100 customers across three regions, and 15 public customer endorsements. The next tier review where this criteria will be evaluated across the network is on October 1. How to find an exam or learn more about the Claude Partner Network Certification preparation courses for each credential are available in the Anthropic Partner Academy. Exams are available to members of the Claude Partner Network. Firms can join the Claude Partner Network and register practitioners at http://claude.com/partners . Membership is free.
Explore more product news and best practices for teams building with Claude.
Transform how your organization operates with Claude
Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.
Please provide your email address if you'd like to receive our monthly developer newsletter. You can unsubscribe at any time.
采集自 Claude Blog，由 collect_claude_blog.py 自动采集
Claude models explained: choosing the best model for your use case¶
来源: Claude Blog 发布日期: Jul 24, 2026 采集时间: 2026-07-25 价值评分: 9.0/10 正文字数: ~4825 字符
摘要¶
How to choose the right Claude model for your workload: compare model classes, weigh cost per task vs. cost per token, and build evals that settle it.
正文内容¶
Our advice: start smart 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 in our documentation on model selection. The Claude model family Mythos / Fable 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. The Mythos class ships in two packages of the same underlying model. Claude Mythos is for trusted organizations handling dual-use cybersecurity and biology work while Claude Fable is packaged with additional safeguards that make the model safe for use by the general public. Both require limited data retention so they can be used safely . Opus 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 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 Haiku is our lowest cost and fastest model class. Haiku models are designed for high-frequency workloads where latency and cost matter. How to choose which Claude model is best for your workload 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 . 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 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.
Explore more product news and best practices for teams building with Claude.
Transform how your organization operates with Claude
Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.
Please provide your email address if you'd like to receive our monthly developer newsletter. You can unsubscribe at any time.
采集自 Claude Blog，由 collect_claude_blog.py 自动采集
Working at the frontier: How Cursor knew Claude Fable 5 was ready for the hardest 1% of problems¶
来源: Claude Blog 发布日期: Jul 17, 2026 采集时间: 2026-07-25 价值评分: 9.0/10 正文字数: ~582 字符
摘要¶
How Anthropic's Claude Fable 5 beat CursorBench and expanded what's possible for Cursor and agentic coding.
正文内容¶
Nate Schmidt's job at Cursor is to evaluate frontier models against their ability to tackle long-running, real-world engineering problems. Here’s why–and how–Claude Fable 5 changed the calculus on what coding agents are capable of.
Explore more product news and best practices for teams building with Claude.
Transform how your organization operates with Claude
Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.
Please provide your email address if you'd like to receive our monthly developer newsletter. You can unsubscribe at any time.
采集自 Claude Blog，由 collect_claude_blog.py 自动采集
Working at the frontier: How Cognition trusts Claude Fable 5 to work through the night¶
来源: Claude Blog 发布日期: Jul 10, 2026 采集时间: 2026-07-25 价值评分: 9.0/10 正文字数: ~549 字符
摘要¶
Cognition tested Claude Fable 5 in Devin, its AI software engineer. It's the first model its team trusts to run unattended for eight hours and deliver production-ready code.
正文内容¶
Silas Alberti, SVP of Research at Cognition, has tested nearly every Claude model inside Devin, the company's AI software engineer. Claude Fable 5 is the first he'd trust to leave running overnight.
Explore more product news and best practices for teams building with Claude.
Transform how your organization operates with Claude
Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.
Please provide your email address if you'd like to receive our monthly developer newsletter. You can unsubscribe at any time.
采集自 Claude Blog，由 collect_claude_blog.py 自动采集
Think through hard problems in voice mode¶
来源: Claude Blog 发布日期: Jul 23, 2026 采集时间: 2026-07-25 价值评分: 8.2/10 正文字数: ~520 字符
摘要¶
Starting today, voice mode runs on Anthropic's Claude Opus, Claude Sonnet, and Claude Haiku, reaches the tools you’ve connected, and speaks many more languages.
正文内容¶
Starting today, voice mode runs on Claude Opus, Claude Sonnet, and Claude Haiku, reaches the tools you’ve connected like Gmail and Slack, and speaks many more languages.
Explore more product news and best practices for teams building with Claude.
Transform how your organization operates with Claude
Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.
Please provide your email address if you'd like to receive our monthly developer newsletter. You can unsubscribe at any time.
采集自 Claude Blog，由 collect_claude_blog.py 自动采集